Skip to main content

Full text of "charlesRiverAssoc :: US vs IBM Exhibit 14971 part 2 Jul80"

See other formats


[7/22/80] 



Historical 


. Narrative 


The 


1960s 


Table of 


Contents 



Page 

IV. IBM's System/360 and the 1960s 269 

34. IBM 269 

a. 3000 Series and the Spread Committee . . . 2T] 

b. The Spread Repo? ^nd S/360 278 

(i) Price/Pe: orn . 2 and Techno- 
logical Leade .^ip — Generally .... 279 

(ii) System/360 Component Technology . . . 282 

(iii) Single Family for All Applications . . 290 

(iv) System/360 Compatibility 296 

(v) Emulation 304 

(vi) System/360 Software 311 

(vii) System/360 Peripherals .' 314 

(viii) Standard Interface/Modularity ..... 332 

Appendix: Examples of System/360 Uses .... Al 

35. The System/360 Commitment 341 

a. Preparation for Announcement ; 357 

b. 360 's Success and Impact on IBM . . - v ... 367 

36. Initial Competitive Responses to System/360 . . 376 

a . RCA 377 

b. GE . . . .' 379 

c. CDC 381 



Page 

d. Sperry Rand 382 

e. Burroughs " . . 383 

f. Honeywell 383 

g. SDS 383 

37. IBM's Responses (1964-66) ..... 385 

a. Reduction of Extra Shift Usage Charges . . 385 

• b. Memory Improvements 386 

c. Tape Drive Improvements 390 

d. Disk Drive Improvements 39 3 

e. Introduction of the Model 20 . . . . . . . 395 

f . The Model 90 Program 401 

g. The 360/44 412 

h. The 360/67 . . 417 

38. Educational Allowances ..... 437 

IBM's Educational Support Programs .... 444 

39. IBM's Unbundling 451 

a. Introduction . . ■. -. 451 

b. The Continued Demand For Bundling ' 

in the 1960s 458 

c. IBM's Unbundling Announcement . 462 

40. Sperry Rand/Univac 473 

a. Univac's Problems in 1964 473 

b. The 1108 477 

c. The Product Line Task Force 480 



Page 

d. The 9000 Series 482 

e. Univac's Success in the Late 19 60s .... 486 

41. General Electric ............ ."V*7 488 

a. The GE 400 Series ' 490 

b. The GE 600 Series 493 

c. Time Sharing 505 

d. False Starts 512 

e. The Management of GE's Computer Operation . 516 

f.- GE's Position in the Late Sixties 523 

h. The Ventures Task Force and the Decision 

to Disengage „ 533 

i. Did GE Lose Money 543 

42. RCA 547 

a. RCA 3301 .' . 547 

b. The Spectra 70 Series 551 

(i) Compatibility with System/360 .... 552 

(ii) Commercial and Scientific Ability •* 

of Spectra 70 Series 558 

"(iii) Problems with Spectra 70 Series . .; . 560 * 

(iv) RCA Success with the Spectra 70 Series 575 

c. RCA Computer Systems Division 1969-1971 . . 578 

(i) Changes in Management Personnel 

and goals 57 8 

(ii) The RCA Series 583 

(iii) Computer Systems Division's 

Problems — Early 1970s 595 

d. RCA's Decision To Sell Its Computer 

Business To Sperry Rand 606 



Page 

e. After the Sale to Sperry Rand 615 

(i) RCA's Activities 615 

(ii) Sperry Rand's Success with RCA's 

Computer Systems Division 616 

f. Conclusion 616 

43. Honeywell 619 

a. The 200 Series . . . 619 

b. Problems and Solutions 626 

(i) Other Systems 626 

(ii) Peripherals . . 62 8 

c. Marketing Practices . 630 

d. Product and Service Acquisitions 

and Expansion 632 

44. Burroughs 638 

a. Burroughs in 19 64: Problems and Changes .. 638 
(i) Reduction of Expenses 640 

(ii) Increased Product Development .... 642 

b. Computer Development 1964-1969 643 

(i) The 500 Systems Family • . 644 

(ii) Smaller Computers 650 

(iii) Peripherals 653 

c. Burroughs at the End of the 1960s 654 

45. National Cash Register 657 

46. Control Data Corporation (CDC) 670 

a. CDC's 6000 and 3000 Series Offering 

(1963-1969) 671 

(i) The 6000 Series 671 



*(ii) The 3000 Series 677 

b. CDC's Expansion into Commercial 

Data Processing 678 

c. CDC's Expanding Peripheral Business .... 682 

d. Data Centers 684 

e. CDC's Acquisitions (1963-1969) 686 

f. Conclusion 690 

47. SDS 691 

a. The SDS Entry Strategy 69 3 

b. The SDS 910 695 

c. The Expansion of the SDS 900 Series .... 696 

d. The Sigma Series 703 

e. The Merger 711 

48. Digital Equipment Corporation 713 

a. PDP 1, 4, 5 and 7 71S 

b. PDP 6 721 

c. PDP 8 722 

d. PDP 10 : . 727 

e. PDP 15 729 

f . PDP 11 729 

g. Peripherals and Software 7 31 

h. Competition 732 

49. AT&T . . . 736 

Competition for Bell System EDP Business . 737 

Other AT&T Competition 746 



Page 

50. The Emergence of IBM Plug-Compatible 
Manufacturer (PCM) Competition 750 

a. From OEM to PCM 753 

b. PCM Entrants 762 

(i) Telex . . 762 

(ii) Ampex 765 

(iii) Memorex 767 

(iv) ISS 775 

(v) CalComp/Centu? ^ata Systems 776 

(vi) Sanders Assr j.at... . Inc 777 

c. PCM Price Competition and Success . . . . . 780 

51. Leasing Companies 797 

a. An Overview of Leasing Company Operations . 79 7 

(i) The Investment Tax Credit ...... 800 

(ii) Marketing Costs 801 

(iii) Capital Availability and Cost .... 803 

b. The History of Leasing Companies in 

the 1960s . 805 

(i) Leasing Company Growth 80 7 

(ii) The Emergence of New Challenges . . . 814 

(iii) Diversification 818 

c. Some Individual Companies 819 

(i) Greyhound 819 

(ii) Boothe Computer Corporation 821 

(iii) Itel 824 



Page 

d. • The Effects of Leasing Companies on IBM . . 826 

52. Service Bureaus " 831 

a. Entry and Growth 831 

b. Time Sharing and the "Computer Utility" . . 835 

c. Competition 840 

53. Software Companies 851 

54. The Role of the Federal Government 866 

55. Planning for New Products 878 

Introduction 878 

Tape Drive Development: The 24 20 and 

3420 (Aspen) ....'. 885 

c. Disk Drive Developments: The 330 (Merlin), 

2319 and 3340 (Winchester) . 893 

(i) The 3330 (Merlin) 893 

(ii) The 2319 and 3340 (Winchester) .... 902 

d. New Processor Planning (NS and System/3) . 907 

(i) Monolithic Logic and Memory 907 

(ii) System/3 911 

(iii) Virtual Memory 913 

(iv) NS Prices . . . . 918 

e. Conclusion 922 

56. Growth of the EDP Industry 923 

a. Increase in the Number of Users of Computers 926 

b. Expansion by Existing Computer Users . . . 928 

c. Explosion of New Applications of Computers. 929 

d. Improved Price/Performance and Ease of Use. 937 

57. Conclusion 944 



L ^ V. IBM'S SYSTEM/360 AND THE 1960S 

2 ! 34. IBM. The 1960s for IBM was an era of great change, 

3 : of great risk and difficulty and most of all a decade marked by the 
4 | phenomenal success of IBM's 'System/360. * The 360 story begins in 



u 
IS 
19 
20 



6; * Several witnesses actively participated in the planning, develop- 
ment and execution of the System/360 program. Their testimony 

7 , provides us with a useful means of understanding this critical 
period in IBM's history. 

8 

Erich Bloch was the engineering ^manager of Project STRETCH from 
g. \ October 19 58 to April 1961, and "was responsible for the circuit 

design and systems organization and implementation". (E. Bloch, Tr. 
[q \ 91468.) In April 1961, Bloch headed IBM's Advanced Technology Study 

Committee, which was established to recommend the appropriate logic 
H I component technology for future products. (E. Bloch, Tr. 9149 2.) 

From June 1961 to September 1964, Bloch led IBM's development of 
[2 i Solid Logic Technology and "was responsible for the development, 

design and pilot manufacturing of the SLT family of components and 
[3 || packaging and their manufacturing". (E. Bloch, Tr. 9146>*)9.) 

l^ ; Dr. Frederick P. Brooks, Jr., hired by I3M in 1956 as an engineer, 

I helped to design the architecture of the STRETCH computer. (Brooks, 
15 | Tr. 22650-51.) In 1960, Brooks became Systems Planning Manager of 
j the Data Systems Division (DSD) and was responsible for developing 
tg ■ "the plans and architecture" for the 8000 series. (3rooks, Tr. 
:| 22656-57, 22665.) Brooks served as Manager of IBM's New Product 
Line project from 1961 until 1964 and was responsible for "think [ing] 
through the technological and architectural approach to a total corpo- 
rate-wide product line". (Brooks, Tr. 22656-57, 22666-67.) From 
early 1964 to the summer of 1965, Brooks was Manager of Operating 
System/360 (Brooks, Tr. 22673-74) and headed the design and develop- 
ment activities for System/3 60 's programming support. (Case, Tr. 
77966-67.) 



7T ■] Richard Case, in 1962, was a member of the Advanced Systems i 

; : Group which was responsible for the design and development of System/ \ 
22 ij 360, and personally headed the engineering group which was at that time! 

j designing what was announced as the IBM System/360 Model 60. (Case, j 

-« ij Tr. 7 2010, 73235-3 8.) During this time frame, Case also served on ■ 

'j IBM's Architecture Committee (Case, Tr. 72008-09; DX 3538), which ! 

2, ;i was responsible for developing System/36 0' s architecture. (Case, Tr . j 

\\ 72008-09.) Case's function on the Committee was to represent all of ! 
2* j the 360 engineering groups. (Case, Tr. 72012, 73238.) In 1964 to 
~ ; | 1965, Case was Assistant Manager of OS/360 (Case, Tr. 77966-67; 

i DX 3538) and assisted Dr. Frederick Brooks (Manager of OS/360) in the I 

i 

'I -269- \ 



L i^ V. IBM'S SYSTEM/360 AND THE 1960S 

2 ! 34. IBM. The 1960s for IBM was an era of great change, 

3 : of great risk and difficulty and most of all a decade marked by the 
4 | phenomenal success of IBM's 'System/360. * The 360 story begins in 



u 
IS 
19 
20 



6; * Several witnesses actively participated in the planning, develop- 
ment and execution of the System/360 program. Their testimony 

7 , provides us with a useful means of understanding this critical 
period in IBM's history. 

8 

Erich Bloch was the engineering ^manager of Project STRETCH from 
g. \ October 19 58 to April 1961, and "was responsible for the circuit 

design and systems organization and implementation". (E. Bloch, Tr. 
[q \ 91468.) In April 1961, Bloch headed IBM's Advanced Technology Study 

Committee, which was established to recommend the appropriate logic 
H I component technology for future products. (E. Bloch, Tr. 9149 2.) 

From June 1961 to September 1964, Bloch led IBM's development of 
[2 i Solid Logic Technology and "was responsible for the development, 

design and pilot manufacturing of the SLT family of components and 
[3 || packaging and their manufacturing". (E. Bloch, Tr. 9146>*)9.) 

l^ ; Dr. Frederick P. Brooks, Jr., hired by I3M in 1956 as an engineer, 

I helped to design the architecture of the STRETCH computer. (Brooks, 
15 | Tr. 22650-51.) In 1960, Brooks became Systems Planning Manager of 
j the Data Systems Division (DSD) and was responsible for developing 
tg ■ "the plans and architecture" for the 8000 series. (3rooks, Tr. 
:| 22656-57, 22665.) Brooks served as Manager of IBM's New Product 
Line project from 1961 until 1964 and was responsible for "think [ing] 
through the technological and architectural approach to a total corpo- 
rate-wide product line". (Brooks, Tr. 22656-57, 22666-67.) From 
early 1964 to the summer of 1965, Brooks was Manager of Operating 
System/360 (Brooks, Tr. 22673-74) and headed the design and develop- 
ment activities for System/3 60 's programming support. (Case, Tr. 
77966-67.) 



7T ■] Richard Case, in 1962, was a member of the Advanced Systems i 

; : Group which was responsible for the design and development of System/ \ 
22 ij 360, and personally headed the engineering group which was at that time! 

j designing what was announced as the IBM System/360 Model 60. (Case, j 

-« ij Tr. 7 2010, 73235-3 8.) During this time frame, Case also served on ■ 

'j IBM's Architecture Committee (Case, Tr. 72008-09; DX 3538), which ! 

2, ;i was responsible for developing System/36 0' s architecture. (Case, Tr . j 

\\ 72008-09.) Case's function on the Committee was to represent all of ! 
2* j the 360 engineering groups. (Case, Tr. 72012, 73238.) In 1964 to 
~ ; | 1965, Case was Assistant Manager of OS/360 (Case, Tr. 77966-67; 

i DX 3538) and assisted Dr. Frederick Brooks (Manager of OS/360) in the I 

i 

'I -269- \ 



I j; the 1960-61 time period. As we have seen, by that time IBM 
o iiwas marketing more than 15 different processors and at least 
3 Ij seven separate lines of second generation computer systems. 

(See above, pp. 126-49.) The architecture of those systems was 
"quite dissimilar", as was their programming. (DX 4740, Evans, 
p. 3925.) Whatever software compatibility there was existed 

y ijonly over a very narrow range of processor performance. 

il 

si! 



20 



9 |! design and development of System/360 ' s programming support. (Case, 
|| Tr. 77966-67, see also Tr. 77977, 77982.) Case was a co-inventor on 

10 j| the original System/360 patent. (Case, Tr. 71993-99; DX 3538.) 

11 i Bob 0. Evans was Director of Systems Development and Planning 

i for DSD from early 1961 to mid-1962 (Evans, Tr. 101269; DX 8031 (Tr. 

12 |j 101035)), and initially responsible for "personally evaluating ... 
|j the 8000 series" and deciding whether IBM should pursue the project. 

13 I (Evans, Tr. 101269.) During this period, Evans also served as Vice- 

j| Chairman of the SPREAD Committee. * (Evans, Tr. 101270; DX 140 4A, p. 3 

14 i| (App. A to JX 38).) In the 1962 to 1965 time frame, Evans was Vice- 
|j President of Development for DSD (DX 8081 (Tr. 101035)), and assumed 

15 || "worldwide responsibility for coordination of the development" of 
!| System/360. Evans, Tr. 101061.) 

16 ji 

I! Paul W. Knaplund was Manager of Systems Marketing for the Data 

17 :j Processing Division (DPD) in 1960 and was responsible for understand- 
ing and informing IBM's product divisions of "the functions and 

18 ;| prices necessary for IBM products to be economically attractive to 

l! users" . (Knaplund, Tr. 90467, see also DX 9033 (Tr. 90458).) In the 

19 ij latter part of I960, Knaplund became Manager of Systems Development 
for the General Products Division (GPD) and was responsible for pro- 
jecting profitability for and meeting profitability objectives of 

.various IBM products, including the 1401 and 16 20 processors and IBM 

21 |j unit record equipment. (Knaplund, Tr. 90464-68; see also DX 9033 

!j (Tr. 90458).) In 1963 Knaplund was named Assistant Group Executive, 

22 j Product and Profit Planning for the Data Processing Product Group DPD 

ij (Knaplund, Tr. 90474; DX 9033 (Tr. 90458)), and "was directly involved: 

23 ; | in the preparations and discussions that resulted in the System 360 
I announcement of April 7, 1964". (Knaplund, Tr . 90474-75.) In the 

24 ; j 196 4 to 196 6 time frame, Knaplund assumed further executive respcnsi- ■ 
■Ibilities as Vice President and Group Executive of the Data Processing . 

25 ij Product Group, and as Vice President and Group Executive of the 

ji Systems Development and Manufacturing Group, which required him "to 

iideal with functional, pricing and schedule issues" relating to System 

!! 36 and other oroducts. (Knaolund, Tr. 90468, see also DX 9033 
1 (Tr. 90453; . ) 

; i -270- 



In addition, the input/output equipment had been developed 
"almost uniquely" for each processor in order to optimize the perfor- 



3 jimance of each of the different system types. (DX 4740: Evans, Tr. 



4 

5 

6 

7 

8 

9 

10 

11 

12 

13 



(Telex) 3925.) The result, of course, was a very limited flexibility 

in attachment possibilities. As Evans testified, because peripheral 

equipment differed for different families or attached in different 

[ways to different processors, customers "had great difficulty in mov- 
i 

ing even from one member of a processor in one family to another, let 
alone moving from one family type to another". (DX 4740: Evans, Tr. 
(Telex) 3925-26.) In this regard IBM's computer systems were no dif- 
ferent from the computer systems of its competitors - (See pp. 15 6-70, 
203-11, 229-37, above.) The result of this situation was that cus- 
tomers generally acquired set systems and had very little flexibility 

14 jito change their configurations as business demands changed. 

\\ 

15 l! a. 3000 Series and the SPREAD Committee. In the early 

il 

16 111960s, IBM's General Products Division (GPD) was resDonsible for the 

jj 

17 Ijdevelopment and manufacture of IBM's small and intermediate systems, 

ii 

18 I'such as the 1401 and 1620, as well as IBM's disk drives. (Knaplud, 

ij 

19 l|Tr. 90464-65; DX 13890, pp. 16, 18; see also DX 1404A, p. 10 (App. A 

20 | to JX 38).) IBM's Data Systems Division was developing and manufacture 
(I 

21 lling IBM's larger systems, the 7000 Series, as well as IBM's tape drives 

22 ij (DX 4740: Evans, p. 3919; DX 13890, p. 16; see also DX 1404A, p. 10 

il 

23 n(App. A to JX 38).) DSD and GPD were achieving great success in the 

24 ^marketplace with their current lin^s — particularly with the 7090 and 

25 11401. (DX 1404A, pp. 81-82, 35, 86, 39 (App. A to JX 38).) In fact, 

-271- 



1 ! 
i 
i 

2 ! 

i 

3 ! 
4 

5 
6 
7 



3 

9 
10 
11 
12 
13 
14 
15 
16 



the 1401, which had been announced in October 1959, was the most suc- 
cessful computer system that IBM had ever introduced, with domestic 
shipments of more than 1600 by year end 1961. (DX 140 4A, p. 75 
(App. A to JX 38); DX 2609B, p. 94.)* 

Nevertheless, neither of the Divisions was resting on its 
laurels; they were planning for the future. If IBM was to continue 
to compete successfully, it would have to commit itself to the devel- 
opment of even better products . Such a commitment would require large 
financial investments by IBM. T. J. Watson, Jr., IBM's Chairman, 
fully understood this requirement and reported the following to IBM's 
Managers in an April 24, 1961, Management Briefing: 

"[0]ur competition is getting stiff er all the time .... 
The best way to meet this competition is to keep our prices 
competitive. Prices involve costs and earnings .... We need 
constantly to spend large sums in research and development of 
new products which will not produce revenue for some years to 
come. Without funds for this vital expense, competition would 
eventually surpass IBM." (DX 8886, p. 43.) 



Thus, within both divisions, improvements and extensions to 
17 lithe then current product lines were being developed. At GPD, Engineer 1 
IS iling Manager Ernest S. Hughes, Jr. (DX 1399, p. 2 (Tr. 33869)), had 



19 j 

20 I 

21 ! 



set up two groups of engineers — one to pursue improvements to the 
1400 family and another to outline and define a replacement for the 
1400 family. (Hughes, Tr. 33915.) At DSD, development was even 

22 !i further alone. A machine called the 810 6 had been under design for 
ii 

23 iisome years and was already under construction within DSD when 

24 ii 

ji 

25 I * w e are aware that DX 26 09B is not in evidence but we rely on it 
I! because it represents a sworn response by an IBM executive which 

j! reflects information taken from IBM's accounting books and records, 
ii 

i -272- 



1 i 
L 



M 

si 

•t 
i 



Dr. Frederick P. Brooks, Jr., came to the Division in 1960 as Systems 
Planning Manager. Thereafter, IBM began to develop the 8106* into a 
series of machines called the 3000 Series. (Brooks, Tr. 22771.) By 
1961, IBM: had spent many millions of dollars en the 8000 Series 
development. (Evans, Tr. 1010 47.) 

Despite the relatively advanced state of the 8000 project 
and the money IBM had already invested in it, there was "vigorous 
debate" within the company over whether the 8000 was the right way to 
proceed. (Brooks, Tr. 22665-66.) With the first elements of the 8000 
nearing announcement, B. 0. Evans, who at than time was Director of 
Systems Development and Planning for DSD, was charged with evaluating 
the 8000 to determine whether it was a "leadership" program. (Evans, 
Tr. 101045-46, 101269.) Evans was charged by DSD's Group Executive, 
T. V. Learson, to get the 8000 into production if it was the right 
thing to do or, if Evans thought the 8000 Series was the wrong 
approach, to do what was right. (Evans, Tr. 101046.) Evans concluded 
that the 8000 Series was "wrong" for a variety of reasons: 

One , the family was based on "contemporary transistor tech- 
nology" and would not be "far-reaching enough". In Evans' view, 
it would have been a "terrible mistake" to build a new family of 
machines that could be rendered obsolete by competitive products- 
incorporating much better transistor technology that would soon 
be available. (Evans, Tr. 101043; see also DX 4773, p. 3.) 



' A 



* The 3106 was an outgrowth of the STRETCH program. (3rooks, Tr 



15 ;22771.) 



-273- 



1 
z 

3 

5" 
S 



Two, the 8000 had a "lackluster" plan with respect to 
peripherals. (Evans, Tr. 101048-49.) 

Three , the 8000 Series was planned to be "a range of five 
different machines: a small scientific machine, a small business 
machine, a medium to high performance business machine, a higher 
speed scientific machine . . . [a] superspeed scientific machine." 



j lj (Evans, Tr. 101051.) Evans thought that offering this "collec- 



3 

9- 
10 

11 
12 
13 

14- 
Is 
16 . 



tion of differing machines with kind of loose ties ... in 
their structure" was "a basic mistake from the user's standpoint". 
(Evans, Tr. 101049.) 
Although Evans believed that the 8000 Series would be an improvement 
over IBM's existing product line and might give IBM a "momentary 
advantage" over competition, he recommended its cancellation. (Evans, 
Tr. 101049; see also DX 4773, p. 1.) On June 27, 1961, W. 3. McWhirtei 
wrote Lear son that I3M's Regional Managers had been apprised of the 
reason why the 8000 was withdrawn: 



l| "[T]he 8000 Series offered insufficient advances to insure - 

^ •] our competitive position at this time — [it] is being replanned 

with new technology to provide a major breakthrough . . . ." 
(DX 14059.) 



13 ;i 

*- ji In late 1961, T. V. Lear son, rhen IBM Vice President and 

22 i| Group Executive, appointed a task force called the SPREAD Committee 
;i 

21 j to develop a new plan for IBM's data processing products during the 

22 ;| 1960s. (JX 38, p. 2; see DX 1404A, p. 7 (App. A to JX 38).) Its 

-- II 

" \ Chairman was J. W. Haanstra, Vice President of Development for GPD 

24 j and its Vice Chairman was Evans, who had become Director of Systems 

^c 1 

-~ ; Development and Planning for DSD. (DX 1404A, p. 3 (App. A to JX 38); 

i 

'j 
:i -274- 



z 
I 

2 

7 . 



DX 3081 (Tr. 101035).) Other members of the SPREAD Committee included 
Dr. Brooks and J. W. Fairclough, Manager of Product Development at 
IBM's Hursley Laboratory in England,* who had been in charge of yet 

Ulf another processor development, the SCAMP.** (Tr. 71984-85; DX 4779.) 
The Committee issued a report of its recommendations in December 
1961. (JX 38, p. 2.) That report is Exhibit A to the System/360 
Stipulation of Fact (JX 38) and is also Defendant's Exhibit 1404A- 

The SPREAD Committee recommended "termination of the prolif- 
eration of IBM products and the development of a family of compatible 
processors which would employ a common technology (Solid Logic 
Technology or SLT) , a compatible set of peripherals and a compatible 
program operating system". (JX 38, p. 2.) The report and recom- 

- || mendations of the SPREAD Committee were accepted by IBM management 



o ■« 



C ! 



* The remaining members of the SPREAD Committee included D. T. 
Spaulding, Group Director of Product Line for the Data Processing 
Group; J. D. Aron, Programming and Technology Coordination Manager 
for FSD; W. P. Heising, Programming Systems Planning Manager, Develop- 
I ment for DSD; H. Hellerman, staff member, IBM Research; W. H. Johnson,- 
i Director of Product Evaluation, Corporate Headquarters; M. J. Kelly, 
j Senior Engineer and Technical Advisor for GPD; D. V. Newton, Manager, 
i Mathematics and Programming for DSD; B. G. Oldfield, Manager, Systems 
I Development for FSD; S. A. Rosen, Data Processing Manager for the 
j Queens IBM New York Branch Office; and J. Svigals , Manager, Systems 
! Marketing for DPD. (Tr, 71984-85; DX 1404A, p. 3 (App. A to JX 38).) 

i ** SCAMP was an experimental computer built in 1960 by IBM's Hursley 
- j Laboratory in England. (JX 38, p. 5.) SCAMP ' s control function was 

|j implemented by the technique of microprogramming. ( Id. ) Although 
2"! the SCAMP project was cancelled in favor of System/360 (DX 4779, pp. 

i| 2-3) , Fairclough was able to convince the SPREAD Committee of the 
3 j benefits of microprogramming, which the Committee adopted as the 

j principal means of implementing control functions in System/360. In 
A "\ addition, because microprogramming techniques were better known in 

;| IBM's British lab than in the United States labs at that time, design 
'5 j of the Model 40 — which was the lead System/360 model in development — 

| was assigned to Hursley. (Brooks, Tr. 22806-07.) (The importance of 

1 microprogramming is discussed below at pp. 3 02-03.) 

I 

1 -275- 



L ! and the development of the New Product Line (NPL) , which ultimately 
2| became System/360,* began in 1962. (JX 38, p. 3.) 

3' ! The principal alternative course of action, which the 

i 

4-1 SPREAD Committee considered and rejected, was the addition of improved 
5 !■ successors to the then existing product lines, rather than develop- 

i 
I 

Si ment of an entirely new line. (Case, Tr» 73571.) The one course of 

i 
7 i action that I3M could not afford to take was simply to maintain the 

j 
3 j status quo and continue marketing its current products . That much 

g. I was plain from the "product survival charts" incorporated in the 

IQ ! SPREAD Report. (DX 1404A, pp. 73-91 (App. A to JX 38).) Those 

i 

t , 

11 j charts "showed that all of the existing products in the IBM product 
£2 j line were estimated to have very short lives, that they would be very 



13 i 



quickly coming out of users' installations . . . [b]ecause other 



14 ; systems manufacturers were developing new and better products and 



that the evaluation was that all of the existing product line was 
very rapidly heading toward being non-competitive". (Evans, Tr. 
101271-72.) 

According to Paul Knaplund, the "'Product Survival Charts' 



15 
IS 
17 
IS 

is! 

•i 
2g :! * The processors included in the April 7, 1964, System/360 announce- 
" I ment and their NPL designations are set forth below: 



21 1 
;i 

22 :i 

■t 

:J 

24 i 

i 



System/360 NPL Designation 

2030 . 101 

2040 250 

2050 315 

2060 and 2062 400 

i 

! 

| 2070 501 (JX 38, <f 4, p. 3.) 

I -276- 



i 

2 ; 

3J 
4-i 

fi 
i 

i 
T\ 

a 

II 

12 S 
13 



. .. . projected displacement of both announced and planned to be 
announced IBM central processing units (CPU's) by newer products as 
users ' needs grew and changed and as new technologies and equipment 
features enabled electronic data processing (EDP) suppliers to offer 
improved products . Those charts demonstrated . . . that IBM had to 
move rapidly ahead with the development, of a new line of CPU's or 
else competition would soon displace IBM's EDP business". (Knaplund, 
Tr. 90473.) 

The survival chart for the 1401 (DX 1404A, p. 75 (App. A to 
JX 38)) made the point graphically. This most successful of IBM's 
systems, announced only two years earlier, was projected to reach a 
peak of installations by 1965 , with installations declining rapidly 
thereafter. Projections for the rest of the product line were similar. 
t ^; l The charts indicated that if IBM did not introduce new, improved 
7 = ! products its entire installed base would be replaced by its competitors 
,g j| In the face of these projections, the SPREAD Committee stated the 
1T :j need for new products to be developed and delivered by 1965. Accord- - 
j- : | ingly, they recommended announcement of the first processors in the 

'i 

, '! line during the first quarter of 1964. (DX 1404A, p. 57 (App. A to 

„ J) JX 38) .) 

« T j The SPREAD Report, and the Systems Architecture Group which 

;! was responsible for implementing its recommendations, created a 

„ :i product plan that went far beyond the recognized competitive need for 

1 
_ .] new and improved products and set forth a revolutionary concept of a 

5C . ; future product family. This concept represented a commitment to a 

i 

| vision of the future development of the industry far more daring and 

i 

t 

i -277- 



/ i 

3 

g 

ia 

12 

13 

L4- 



LI far-reaching than any of IBM's competitors ever attempted. The 
Z l concept, subsequently embodied in IBM's System/360, held the potential, 
3. j if successfully implemented, for enormous business success for I3M 
4-'r and also for revolutionizing the EDP industry. It sought not just 
competitive success with existing users but a vast expansion of the 
number and. types of EDP users and uses . At the same time , the magni- 
tude of the commitment — the devotion of virtually the entire business 
to that concept — carried with it a risk of staggering proportions . 
Both internally and externally, the IBM System/360 program came to be 
referred to as a "you- bet your company" undertaking. (Evans, Tr. 
101126; see also Friedman, Tr. 50 373; Case, Tr. 73561-62.) 

b. The SPREAD Report and S/360. The concept for the New 
Product Line (NPL) , which became 360, embodied a number of objectives 
including : 

(i) the clear assertion of price/performance 
and technological leadership; 
:j (ii) the merger of business and scientific capabilities 

17 il 

I in a single family of systems (in fact, the attainment of a 

iS ;l 

ij series of computer systems that would be an industry leader 

13 ; i 

! in the performance of all applications , hence the origin of the 

i name "System/360" to denote the full 360 degrees of the circle 
21 \ 

] (Evans, Tr. 101129)); 



(iii) upward and downward compatibility across a broad 
23 "I 

| family of processors; 
21 1 

! (iv) a comprehensive set of systems software; 

M* St 

j (v) compatibility of a wide range of peripherals across 

1 
1 
! -273- 



1 i 

3-j 

4-1 
» 

si 

7 I 

3 

9- 

a 

1 

2 
3 

.4 



the entire family of CPUs ; and 

(vi) the substantial user flexibility attainable from the 
resulting modularity of the boxes constituting a 360 computer 
system. (DX 1404A (App. A to JX 38).) 

Each of the objectives held the promise of greater customer 
acceptance and a substantial broadening of the demand for and use of 
computers , and each raised its own particular challenges and risks . 
The attainment of each objective posed obstacles in development, 
design and manufacturing, each of which carried with it the possi- 
bility of failure. These objectives, and the manner in which IBM 
successfully attained them, are discussed in the following pages. 

(i) Price/Performance and Technological Leadership — 
Generally. The 8000 Series was cancelled because it would not have 



been a "leadership" product for a significant period of time, 
either technologically or in a price/performance sense. System/360, 
its replacement, was clearly intended to be both. In December 196 2, 
•[ T. V. Learson, IBM Vice-President and Group Executive of the Corporate 
I Staff, wrote to Evans that IBM's aim was to make the new line "eco- 
!} nomical as hell, simple to operate and the best on the market". 



.6 
.7 
3 

9 |i 

] (DX 4795.) IBM's Chairman wrote in June 1963 that it was important 

:j for IBM to "make these machines good enough so they will not be just 
II ij 

ij equal to competition" , because IBM expected that once they were 
2. ';[ 

;| announced IBM's competitors would " immediate lv try to better them" 

3 i 

! and "I [Watson] want cur new line to last loucr enough so we do not go 

14. II 

j in the red". (DX 48 06.) 



-279- 



L i 

2 

3 

5 
6 I 
7 

a ; 
9- 

ia 
ii 

12 i 
13 

14 
15 

15 ;i 

'I 

is -i 

19 1 

■f 

2C; ! 

21 :j 

:i 

22 ; 

\ 

23 I 

24 1 



The price/performance of System/360 turned out to be a 
spectacular improvement over IBM's earlier product line. (Rooney, 
Tr. 12123-24; Welke, Tr. 17079-80, 17304-05; see also Northrop, Tr. 
82711; PX 3638 ») In a memorandum written to Evans and Kennard just 
prior to the System/360 announcement/ the Manager of Market Analysis 
for the Data Systems Division stated that " [i]t is difficult to 
estimate the competitive jolt NPL will create. Never before has a 
single announcement obsoleted so much existing equipment at one time" 
since "NPL will have an advantage over all existing systems offered 
by major competitors". (PX 1099A, p. 1.) In particular, the Model 30, 
intended largely to replace the 1401, had "six times greater internal 
speed" than that system. (DX 3726 (Tr. 78990); see also DX 4740: 
Evans, Tr. (Telex) 4034-35; DX4755.) 

The following comparisons at the time of announcement illus- 
trate these improvements: 



Rental Price (with maximum memory) 



Maximum Main Memory Capacity 
(Chars. ) 



Performance 
(instructions/second) 



Performance/Rental Price 



1401 

$2,680 

(DX 573, 
p. 6) 

4,000 
(DX 573, 
p. 3) 



Model 30 

$3,875 
(JX 38, 
p. 33) 

65,536 
(JX 38, 
p. 32) 



5,000 
(DX 4740 : 
Evans, Tr. 



30,000 

(DX 4740 : 
Evans , Tr . | 
. (Telex) 4034-35; (Telex) 4034-i 
DX 4755) 35; DX 4755) j 



1.87 



7.74 



-280- 



2 

3 
4 

^ 
a 

7 ! 

3 

9- 

10 I 

11 

12 

12 ! 

14 

15 

16 

17 ! 

13 ] 

•I 
19 :| 

za } 

21 :| 

lj 

22 "» 

23? 

i 

l 
.i 

24 : i 

:! 
i 

25 



Rental Price (with maximum memory ; 



Maximum main memory capacity- 



Rental per million characters 



Performance (multiplications/second ) 



7090 

$43,500 

(DX 572A, 
p. 6) 



Model 75 

$60,300 
(JX 38, 
p. 394) 



196,608 (6 bit 1,048,576 (8 bit 
characters) characters) 



(DX 572A, 
p. 5) 

$100,708 

(DX 572A, 
p. 6) 

38,200 
(Case, Tr. 
74220) 

.88 



(JX 38, 
p. 394) 

$35,286 
(JX 38, 
p. 394) 

366,000 

(Case, Tr. 
74220) 

6.07 



Performance/Rental Price 

Knaplund testified that just prior to announcement IBM con- 
sidered 360 price/performance to be superior to the best known 
competitive systems and substantially superior to the best of IBM's 
older computer systems. (Knaplund, Tr. 90503; see also PX 1099A.) 
T. V. Learson wrote in July 1964 that System/360 's price/performance 
had achieved a 30% to 50% improvement over IBM's previous produce 
line. (DX 1525.) Moreover, the analytical methods used at the time 
to predict price/performance understated the comparative advantages 
of System/360 by failing to take into account the benefits to the user | 

stemming from the use of disks, the advantages of compatibility, the ! 

i 
System's improved reliability, the advantages expected to come from its! 

software and the availability of large memories. (Knaplund, Tr. 90504- | 

05; see also PX 1099-A, p. 1; PX 6204, p. 1.) ; 

i 

The price/performance and other advantages of System/3 6 j 



-281- 



i ; 

i 

z\ 

a ! 

i 

4-j 

si 

i 

I 
s i 

7! 
3 i 

si 

10 ! 

I 
I 

12. | 
13 ! 



were recognized outside of IBM as well. For example, plaintiff's wit- 
ness Frederic G. Withington of Arthur D. Little reported in October 
1964 that " [w] ith the introduction of their System 360 equipment, IBM 
established the new price-performance standard, for equipment within 
the computer industry for the next several years" (PX 4829, p. 16) r 
an. opinion which he reiterated during his testimony. (Withington, 
Tr. 56591-92.) Withington also testified that System/360, at the 
time of its announcement in 1964, was "regarded as the best in terms 
of incorporating the most recent developments in systems program and 
machine architecture".. (Tr. 56590.) Similarly, in a June 1964 
presentation to G. E.'s Executive Office, John Weil called System/360 
an "excellent product line with outstanding peripheral offerings" and 
stated that it was "no longer possible to offer equipment with a 



14 I* significant advantage over IBM". (PX 320, pp. 13-14.) Additionally, 



15 I 

1 

16.1 
IT J 

is ; 
19 ; 



RCA's June 1964 "Five Year Plan" noted that System/360 "has and will 
have a significant impact on the marketplace and other suppliers are 
obliged to meet its capabilities". (PX 243, pp. 5-6.) 

(ii) System/360 Component Technology. In explaining his 
recommendation to cancel the 8 000 Series, Evans had written: "New 



m ~ '1 technology is essential to a new IBM machine family. Committing a 

:! 
„ T ;j new family's lot to current technology is opening IBM to a major 

•! competitive coup". (DX 4773, p. 3.) The improvements in price/ 

„ ij performance offered bv 360 could not have been achieved without the 
! 

,| superior circuit technology that Evans had envisioned. (Case, Tr. 



-282- 



I- i 73244; Evans, Tr. 101048.) Development of such technology (called 

2. | Solid Logic Technology or SLT) had already begun in IBM when the 

31 ;. SPREAD Committee met. The Committee recommended the use of SLT as 

4-1 processor componentry because it "promised improved cost/performance 

SJ- and reliability." (JX 38, p. 5; DX 1404A, p. 7 (App. A to JX 38).) 

S ; Case testified that "the entire System/360 line . . . was predicated. 

7; on the availability of the new SLT technology." (Case, Tr. 72303.)* 

8. ; SLT development, which had begun prior to 1961, was acceler- 

g ; ated in April of that year on the recommendation of IBM's Advanced 

i 
1XJ ! Technology Study Committee that a "high priority SLT program" be 

JX ■ established^ (JX 38, p. 5.) According to Erich Bloch, who headed 

12, \ the Advanced Technology Study Committee until September 1964, that 

J3 ; Committee had been charged with recommending the logic component 
i 

L4 ; technology that IBM should use in its future EDP equipment and with 

£5 ■; establishing the schedule and cost objectives for its implementation. 

1$ jj (E. Bloch, Tr. 91468-69, 91492, 91686; see also DX 9117, p. 2.) 

The Committee decided that the new technology had to be 

producible at half the cost of the then current SMS (Standard Modular 

^g ■! System) technology and be four times as fast. (E. Bloch, Tr. 91492- 

2g it 93.) These performance goals were influenced by both the technology 

ij 
2«r ;| performance and computer performance that could be achieved by IBM 

2^ .] competitors , including both computer manufacturers and component 

!i 

• I 

23 •! 



17 
18 



-a '! * "In fact, in its early states, the Svstem/3 6 oroject was known 
Vjas the SLT family." (Case, Tr. 72303.) 



?« 



-283- 



1 

2 
3 

5 
5 
7 
3 
3 

ia 
ii 

12 



L5 



manufacturers. According to Bloch, the existence of such competitors 
and their introduction of new products and processes since the 196 0s 
has forced IBM to be alert to their offerings in order to remain 
competitive in terms of cost, performance, reliability and function. 
(Tr. 91690-9-2.)* As T. V. Learson later put it, prior to the intro- 
duction of System/360, "IBM had been in the market for a long time 
with the old circuitry .... It was time for a change. Competitive 
action told us so . . . . " (PX 1900, p. 7.) Because of such 
"competitive pressures", the Committee decided that the development 
of the new technology had to be accomplished within 18 months and the 
delivery of machines incorporating the technology to customers begun 
within three years. (E. Bloch, Tr. 91686-87; see also DX 9117, p. 
3.) 

The Committee considered three courses of action: improve- 
ment of the existing technology; development of monolithic technology; 
*q ;j and further development of a hybrid technology (SLT) with discrete 

T |[ semiconductor components combined with screened circuit elements. 

Q ! (E. 31och, Tr. 91492.) SMS was the packaging for discrete components 

IS ;| 

Tq |i used by IBM in its second generation equipment. (Case, Tr. 72265.) 

^ \ It had been designed and developed by IBM for Project STRETCH and was 

i 

] superior to the discrete component packaging available from outside j 

;j suppliers because it was optimized for use in EDP equipment. (E . Bloch; 



T 



\ * Bloch included such firms as Texas Instruments, Fairchild, 
2 ~ :j Motorola, Intel, Mostek, AMD, Hitachi, Fujitsu, Philips and National 

.[Semiconductor. (S. Bloch, Tr. 91691-92.) 

25 ; j 

i 

-284- 



L 
Z 
3 

4- 



Tr. 91486-87.) Despite its contribution to IBM's successful line of 
second-generation computers, the Committee concluded that SMS tech- 
nology had apparently been pushed close to its limits in terms of 
cost, performance and reliability and would not yield the desired 



5" If performance improvements* (E. Bloch, Tr. 91493; see DX 9117, pp. 4- 

6 i 5, 8.) 

7 j, In order to gain additional information about the feasi- 
3 ; bility of going directly to monolithic circuitry,* IBM was advised by 

g. j other companies, including Fairchild, Texas Instruments and Motorola, 

i 
IQ ! concerning their development activity with monolithic technologies. 

i 
l^\ (E. Bloch, Tr. 91694-95.) The Committee concluded that while mono- 

£2 i lithics could meet the performance requirements laid down they could 

13 || not be produced in the time or at the cost desired. (E. Bloch, Tr. 
i 

^4 j 1 91492-94; see also Gibson, Tr. 22625-27; DX 4782.) The Committee 
75 ; therefore recommended moving part, but not all of the way to mono- 
tg j lithics: the continued development of the "hybrid (SLT) configura- 



17 i 

13 • 



tion". (E. Bloch, Tr. 91492-94.) 

That recommendation was based on several advantages to the 
1Q '! SLT technology: first, it would lend itself well to automation and to 
„g ii a fast production buildup; second, it would lend itself "to a product 
^ :} spectrum of applications" in processors of all sizes as well as 
_,:' input/output devices; third, it would be capable of providing the 

j 
23 -j 

j A ;| * " [T]he total integration of all devices . . . and interconnecting 

"^ ;j wiring in a single piece of semiconductor material". (E. Bloch, 

- c : ;Tr. 91492.) 

1 
i 
■1 

it -285- 



necessary speeds or performance ranges; and finally, the semiconductor 
packaging would accommodate the semiconductor well, provide the needed 
electrical characteristics and give the desired packaging densities - 
(E- Bloch, Tr- 91699-701.) 

These anticipated advantages were in fact realized and SLT 
3 it became a high-performance technology for its day, offering a sub- 
stantial increase in speed at a substantial reduction in size. 
(Case, Tr. 72301-03; E. Bloch, Tr. 91705; see also PX 3587 (Tr. 
25334).) SLT was a "significant advance" in IBM computer technology: 
it required less space, power and cooling per circuit than SMS; it had 
higher performance and "ten times the reliability" of the earlier 
technology — all at a reduced cost. (E. Bloch, Tr. 91496-97; see also 

12 ; | McCarter, Tr. 88380; Evans, Tr. 101132.) Thus,. SLT enabled IBM to 
?iL ' offer "very substantial gains" in price/performance. (Evans, Tr. 

r - ; 101132.) Further, SLT "lent itself to automation" (E. Bloch, Tr. 

j 

,- !•• 91705) and IBM took advantage of that fact by investing heavily in the 
development of automatic tools. (Case, Tr. 72298-301.) "IBM 
coordinated the development of tools, the development of a design 
I automation system and the production and testing of components with 



- ! 

i 

i 

5 \ 
if 

7 

3 

a 

ia : 

ii 



17 
IS 



19 :» 

'] the development of the components themselves. Each of the parts of | 

20" ! ! 

:l the technology took into account the other parts." (E. Bloch, Tr. j 

21 ; | j 

i 91497-98.) | 

22 1 . i 

ij Such automation enabled IBM to reduce 'production costs and ■ 

22 i j 

j improve the reliability of its circuits. (Case, Tr. 72301; E. Bloch, ; 

2^ ! I 

] Tr. 91497.) IBM's "substantial investment" in automatic manufacturing 
2= i ■ 

j techniques was a very important factor in allowing IBM to make System/ i 

'i ! 

i ; 

i ; 

1 -286- I 



360 much more powerful for the same price or to be a lot less costly 
for the same, power. (Case, Tr. 72301.) From 1965 to 1969 5LT tech- 
nology and the automation which accompanied it gave IBM a cost advan- 
tage over other component manufacturers who moved their assembly 
outside the United States in order to get a cheaper labor source for 
the relatively labor-intensive production processes. (E. Bloch, Tr. 
( ;i 91708.) No other computer manufacturer had the equivalent of SLT 



technology at the time of System/360' s announcement and. delivery 
(Evans, Tr. 101131) , despite the substantial benefits that it held 
and despite the fact that SLT was an extension of the existing tran- 
sistor technology which was readily available to everyone. 

Only with the benefit of hindsight, however, was it obvious 
that the SLT decision was the correct one. During the middle 1960s, 
a ; up to about the beginning of 1966, criticism of the decision was 
- i expressed within IBM. Critics thought that SLT had been the wrong 
choice, that by being more aggressive IBM could have gone to mono- 
lithic circuits and taken a larger jump forward.* (E. Bloch, Tr. 
91695-96.) Implicit in that criticism was the apprehension that IBM 
would be the victim of a competitive coup by other companies moving 
n \ beyond I3M in circuit development. 

"! This failed to happen. Based upon a comparison of the cost 

j and capabilities of IBM's SLT circuits with competitive monolithic 



\ 



j i 
r 



> i 
* 



j circuits that became available from the mid-19 60s forward, Bloch 
13 t 



| * 31och also testified that the criticism "died down" when it became 
£ I clear that "SLT met all the goals" set for it in a way that could not 
I have been done with monolithics. (Tr. 91696.) 



-287- 



L 1 
i 

a! 

> 

4-i 

i 

5 

7! 
3 
9 

ia 
11 
12 1 

12 I 

15 ; 

i 

16 ■; 
17 
IS 



concluded that SLT had as good a performance as those later developed 
products, was "much denser" and was produced at lower cost than the 
products which IBM's competitors acquired from outside vendors. 
(E- Bloch, Tr. 91704-05; see also Withington, Tr. 56591.) Moreover, 
when IBM did convert to monolithic circuits in 1968-1970, it was able 
to use a great deal of what had been done in SLT to ease the transi- 
tion into monolithics. (E. Bloch, Tr. 91698; Dunlop, Tr. 93991.) 
This planning for the future had been taken into account by the 
Advanced Technology Committee and for that reason IBM designed tech- 
niques and tools during the SLT development that could be adapted to 
the manufacture of monolithic circuitry. (See E. Bloch, Tr. 91500, 
91703, 91494 and Case, Tr. 72300-14 for details of the carryover of 
SLT development into monolithics.) SLT still is being used by IBM in 
secondary circuit functions of newer products. (E. Bloch, Tr. 91499.) 
The advantages of automation, of taking an intermediate step 
toward monolithics, and of coordinating circuitry, component and 
|[ product development could be fully realized only through in-house 
development and manufacture. Accordingly, the Advanced Technology 



!i Study Committee recommended the establishment of a components division 

20 '" 



19 ;| 

which would be able to manufacture SLT on a large scale. (E. Bloch, 



.„ ] Tr. 91562.) 

21 |j . 

„ ;! Case called IBM's decision to develop and build its own new 

22. * 

i • l 

ij circuitry "perhaps the riskiest single decision that had to be made by i 

■! I 

, IBM in the development of System/360". (Tr. 73514.) It required a | 

24 I I 



?c 



substantial capital investment in a new business — developing and manu- 
facturing transistor components — in which IBM had had little prior 

-288- 



4| 

ff | 

i 

s ! 

7 1 
3 | 

n I 
12 i 

i 
13 ! 



experience. Not surprisingly, there was considerable debate within 
IBM whether components was an "appropriate business" for IBM to get 
3 || into, and the decision to establish the Components Division in 1961 
continued to be second-guessed well into the 1960s — long after IBM 
had committed itself to the point when there was no turning back. 
(Case, Tr. 73515.) In short, as T. V. Learson put it in 1966: IBM 
"had to become, in a very short time, the largest component manufac- 
turer in the world". (PX 1900, p. 9.) If IBM were successful, the 
potential benefits overrode those risks : 

(a) in-house manufacture could help IBM reduce its total 
costs by eliminating middle-man profits; 

(b) by designing the new circuitry and the new machines 
simultaneously, IBM could get the best new circuitry earliest 
because IBM would not have to wait for another firm to finish its 
circuit development process and make the circuit available in 

1g I order to explore the circuit's potential characteristics and use 
in a computer system; 



17 1 
13 ■ 



19 If 



(c) unlike other manufacturers who were less integrated and 
who would have to adapt generalized circuitry to their particular 



;! needs, I3M would be able to enhance the price/performance of its 

;j computer systems by tailoring its own circuitry to the require- 

; i ments of System/360. (Case, Tr. 73245-48; see also E. Bloch, 
22. ,' 

;| Tr. 91563.) 

" i 

■{ In-house manufacture would also oermit IBM to accelerate the training j 
/A s *" i 

.! of computer engineers in both the characteristics and use of the new I 

"i . ! 

.1 circuit technolocy. According to Case, it was believed that IBM "could! 

:i " | 

'I -289- S 



synchronize the development activities between the circuit development 
organizations and the computer development organizations more effec- 
tively if they were in one corporation rather than if they were in 
two or more corporations." (Case, Tr. 73250.) 

Such synchronization was to grow increasingly more impor- 
tant.. Bloch testified that as the integration level of components 
increases, "more and more of a machine is on a single component. 
And therefore when one has in mind the designing of a new computer 
one can learn a lot by just looking at the individual components that 
go into it." (Tr. 91929; see also Case, Tr. 73251-52.) As the 
degree of component integration increased during the 1960s both 
symbiosis in development and confidentiality became increasingly more 
important reasons for in-house development. 

IBM's Advanced Technology Study Committee took the long 

view in 1961. It was building for the future (E. Bloch, Tr. 91929) 

and considered the benefits which might be derived later on from a 

long-term kind of' process worth the risk. (E. Bloch, Tr. 91928.) 

;[ That long range planning paid off handsomely. Case testi- 

.i 

!| fied that IBM achieved the objectives that it set with respect to the 



L 
2 

3. 
4- 

£ 
6 
7 
8 
9- 
10 

11 

j 

12 j 

13 ! 
14- i 

15 | 

16 ! 

17 I 
IS 

19 , 

;! design, development and manufacture of SLT (Tr. 73267) and that 

2G ! 
■I 

'! tne ultimate success of System/360 was "in large measure" dependent 

21 ; i . 

:j on the success of that circuit development activity. (Tr. 73253.) 

22 : i 

|| (iii) Single Family for All Applications. The SPREAD 

22 !j 

.; Committee recommended development of a single line of processors to 
i; "meet the needs of the commercial, scientific, and communications and 

I 
! 

:t 

<\ 

1 -290- 



1 

2 

3 
4 

5 
6 
7 
8 

9 
10 

11 
12 
13 
14 
15 

16 ! 

17 I 

18 j 

19 I 

j 

20 j 

i 

21 i 

22 l 
23 

24 
25 



control markets". (DX 1404A, p. 12 (App. A to JX 38).) That objec- 
tive called for a "fundamental change" in IBM's design emphasis (DX 
4740: Evans, Tr. (Telex) 3925-28) but one which was thought to be 
necessary for developing user requirements* At the time of the 
SPREAD Report, IBM's product lines were "distinctly either commercial 
or scientific in their emphasis". (DX 1404A, p. 13 (App. A to JX 
38).) This was true of other vendors' product lines as well. Up to 
that time, customers who wanted to do what had traditionally been 
considered both "scientific applications" and "business applications" 
generally acquired two computers. (Case, Tr. 73329.) 

By the end of the 1950s, however, the distinctions between 
business and scientific applications were beginning to blur, and 
"customers themselves were not observing [the] lines between scien- 
tific and business machines in actual practice". (Case, Tr. 73274- 
75; see also Tr. 73276-83, see pp. 81-83, 102, 148, 162, 213-15, 239, 
above.) Evans testified that "more and more" often, the "scientific 
side" of a user's operation needed the data handling capabili- 
ties associated with business data processors and the "business side" 
needed the arithmetic and logic capabilities associated with scienti- 
fic systems. (DX 4740: Evans, Tr. (Telex) 3927-28.) The history of 
the 1950s and early 1960s is full of examples of "business" computers 
doing "scientific" applications and vice versa. (See above, pp. 15-21 
38-45, 81-83, 102, 138-49, 162-68, 206-15, 242-44.) 

That user need for "dual use" was a major factor in 
the SPREAD Committee's thinking. According to Evans, "One of 
the premises from the beginning was there would be great 



-291- 



1 

2 

3 

4 

5 
6 
7 
3 
9 
10 

11 ! 
i 

12 | 

l 

13 I 



savings to the users if we could combine in the single machine the 
ability to cover the full range of business applications and scien- 
tific applications as well. So our concept was a single machine that 
would be equally able in either of those areas". (Tr. 101052.) 

Although the Committee foresaw the need for separate develop 
ment of ruggedized products for military purposes , it stated that 
"standard products will satisfy about 32 percent of the available 
military market" and that a basic objective should be "to further 
penetrate the ultra-reliable portion of the military market with the 
SPREAD family". (DX 1404A, p. 44 (App. A to JX 38).) 

Thus, in accordance with these recommendations, it became 

i 
an objective to design the NPL architecture for the "broadest possible} 

range of applications . . .. equally well suited" to what had pre- i 



19 
20 



14 ! viously been considered scientific or business computing. An 

15 j instruction set and processing capabilities were to be designed to 

16 j be "equally suitable to both of those classes of applications and 

;i 

17 jj indeed well suited to the broadest possible range of applications 

ii 

18 |j that one could think of" (Case, Tr. 73268-69), including process 
control applications and communications control applications. (Case, 
Tr. 73321). Evans testified that the name "System/360" was chosen 

21 j for the new line to indicate the "full circle of the applications 
il 

22 jability of the machine". ( Tr. 101129.) 

The combining of capability to do the whole range of applicaj- 

I 
tions in a single machine promised great savings to users and great 

returns to IBM. It was far from clear, however, that the objective 

-292- 



23 i 

24 ! 

25 ! 



z 

3 

5 

5 
7 

a 

10 ; 

| 

n ! 

12 j 

13 

L4 

15 

IS 



of designing "dual purpose" computers could be accomplished without 
a degradation of either performance in business applications or per- 
formance in scientific applications or, indeed, in all the applica- 
tion areas. Evans testified that this risk, was perceived by IBM 
management and "haunted" them.* (Tr. 101052, 101129.) 

"The question was whether we could build machines that in their 
own right as a scientific performer would be the best and also 
had the ability to do the business kind of a problem, or in so 
doing would we really be building mediocrity and someone could 
come along and optimize as the industry had done before and build 
better scientific machines, better business data processors, and 
in the process negate our plans and our aspirations." (Tr. 
101052; see also Case, Tr. 73538-39.) 

The risk that competitors might specialize and, in so doing, 

outdistance a line of products aimed at a wide variety of applications 

was compounded by the risk that, even if 360 was as powerful as more 

specialized competitive machines in their specialties, customers might 

reject System/360 because they just "might not see it that way". 

(Case, Tr. 73538-39.) In the face of these risks, some people in IBM 

: | became proponents of continuing work on the pre-existing "scientific" 

17 I 

l^and "business" product lines. During 1962 and 196 3 there was continued 

13:! 

'I a project to build a scientific computer compatible with and as a 
19 ;! 

;|successor to the 7094 (Brooks, Tr. 22843-44; Case, Tr. 74574); and 

7Q 1 

,t as late as December 1963-January 1964, a group in the General Products 

:! Division led by John Haanstra opposed development of the 3 60/3 in 
•1 
70 :i 



23 



j * Case testified that IBM management "frequently" inquired of the 



.360 design group whether the performance objectives for System/360 
*"" Iwere beinc met for both business and scientific use. (Tr. 73539.) 



,1 



-293- 



1 I favor of extending the 1401 line in its place. (Evans, Tr. 101187-88, 

j 

2 | 101275-76; Hughes, Tr.. 33970-71.) 

3 | Such fears were not unfounded. As we shall see, competitors 

4 j did attempt to offer more specialized systems* to meet the needs of 

i 

5 j certain users and were successful in competing against System/3 60 

i 

6 j where customers wanted such relative specialization rather than the 
j 

7 i more generalized range of functions which System/360 offered. Some of 
i 

8 j the history of the latter part of the 1960s is the history of IBM's 

9 j attempts to respond to such competition. 

1Q. I Despite the risks, the concept proposed by SPREAD was 

11 j pursued. System/ 3 60 was designed to be a machine equally powerful in 

12 | scientific and business applications and with facilities for real- 
i 

13 | time applications, which "machines of that age had not been able to 

14 I address before System/360 with real power and versatility". (Evans, 

15 i Tr. 101144.) 

i 

;i 

16 it Weil testified that the distinction between scientific and 

;j 

17 ;|commercial processing was "erased" " [i]n a practical sense, with the 

ii 

18 -announcement of the IBM 360". (Tr. 7189; see also Beard, Tr. 10342; 

19 [Friedman, Tr. 50378; O'Neill, Tr. 76194-96.)** That testimony was 

20 | 

21 | * The "specialization" offered by competitors was a matter of 
[[degree — many "specialized" competitive offerings could be and were 

22 Ijused to perform a range of applications, but were marketed as more 
•1 "tailored" machines to attract certain users. For example, G.E. 

23 ijinitially targeted its 600 series primarily for engineering and 
;!scientific applications (Weil, Tr . 7026-27) and CDC originally de- 

24 II signed its 6000 series to perform scientific applications. (Norris, 
"Tr. 5617, 5618, 5629; see pp. 423-24, 672-80, 690, below.) 



25 



**" Weil also testified that "Since the early sixties, it really 
j hasn't been economically important to design a computer system only 



-294- 



consistent with Weil's assessment of 360 in June 1964: 

" ► . . System/360 integrates into a single set of equipment the 
capability for business data processing, scientific calculation, 
data communications, and process control. It seems clear that 
all of these are now but facets of the basic information handling 
and processing system.". (PX 320, p. 13.) 

System/360 f s ability to "do the 360 degrees of the circle" resulted 

in acceptance by users who could not get the same range of perfor- 

I ;j mance from other architectures. (Evans, Tr. 101132-33.)* Its broad 



range of applications helped simplify customers' acquisition deci- 
sions, enabled them to achieve economies of scale by acquiring one 
large -capacity, rather than two smaller-capacity, machines and 
permitted them to reduce the required training and improve the 
7 ;j efficiency of their EDP staffs. (Case, Tr. 73327-28; see also Weil, 
Tr. 7059-60.) The combination of business, scientific, and other 
applications in the same line also helped reduce IBM's costs. It 
enabled IBM to concentrate on a single machine type with fewer sets 
of program support and software and with a single set of training and 

'j education for customers and IBM personnel. (Case, Tr. 7332 8-29, 

7 11 
•i see also Tr. 73387-89.) 

3 ;l 
4 Some of the benefits associated with the "erasure" of the 

3 :} 

'! business-scientific distinction and some of the techniques used to 



\ for business or only for scientific applications , except at the 
- S extreme ends of this spectrum, where you are trying to do as much 
^ J scientific calculation as you possibly can within the limits of the 
3 j technology" . (Tr. 7190.) 

! 
i 

4 I 

• j * The diversity of applications to which users applied System/360 
■are described in more detail in the Appendix to this section. 



-295- 



5 

6 

7 

3 

9 
10 
11 
12 ! 
13 
14 
15 
16 



1 ! effect it were also associated with the achievement of another objec- 

2 ! tive of the SPREAD Committee. This was the objective of having a 

3 single compatible line of processors with compatibility extending over 
4- a wide performance range. Compatibility in this sense meant that 

programs written for one processor in the line could be run on a 
second processor, provided that the second processor had at least the 
minimum memory capacity and complement of input/output and auxiliary 
storage devices required by the program, and that successful execution 
of the program did not depend on the speed of the CPU.* (Case, Tr. 
73368-69; see also Brooks, Tr. 22681-82.) 

(iv) System/360 Compatibility. The SPREAD Committee recom- 
mended the development of a new family of compatible processors by 
IBM: 



"IBM customers' needs for general-purpose processors can be most 
profitably met by a single compatible family extending from the 
smallest stored-program core-memory machine to the machine for 
customers growing beyond the 7 94 and 7030. There are proces- 
sor needs above and below this range-it is not yet evident 
that these can be compatible with the new processor family." 
17 i! (DX 1404A, p. 8 (App. A to JX 38).) 



18 
19 
20 I 

i 

22 I 

i 

23 ! 

24 J 

25 ! 



The new family was to consist of at least five CPUs — those five to be 
upward and downward compatible with one another. (DX 1404A, pp. 16, 
25 (App. A to JX 38).) According to Evans, this concept of compati- 
bility envisaged by the SPREAD Committee and implemented in System/360 
was "just a mile apart from the rest of the world". (Tr. 101141.) 



* These three requirements are satisfiable in 90 to 99 percent of 
all the programs that normal businesses execute, according to Case. 
(Tr. 73368-69.) 



-296- 



I ; 

»- : 

v •; 



f 
3 

7 

9 

2 



Prior to the introduction of System/360, it was generally 
true that the computer lines of a particular manufacturer were not 
compatible with one another. (We Ike, Tr. 19193.) Although both IBM 
and a number of its competitors had achieved upward compatibility 
over a "very narrow performance range" covered by two or three machines, 
no one had achieved the full upward and downward compatibility over 
the "very substantial" systems performance range of System/360. 
(Evans, Tr. 101140-41.) Thus, several months after System/360 was 
announced, Withington wrote that "the degree of upward and downward 
compatibility that is achieved with System 360 . . .is certainly by 
far the greatest to date". (PX 4829, pp. 17-13; see also Case, Tr. 
73406-10.) 

The SPREAD Committee viewed compatibility for an entire 
family as a "major advance" that would appeal to customers and "sell 
more processors". (DX 1404A, p. 35 (App. A to JX 38).) From the 
customer's perspective, the Committee regarded compatibility as a 
"powerful selling tool" because it would 

(1) protect his programming investment; 
; i (2) permit phased growth; 

*** ! 
— r 

j (3) minimize his investment in personnel training; 

H 'i 

i (4) expand the available labor market of personnel trained 

!■» ! 

, ' u I 

;| to operate in his environment; 
2. ,| 

j (5) simplifv the adaptation of his applications to several 

! processors; 



— i 



(6) permit him to transfer applications among installations; 



-297- 



1 

2 
3 

4 

5 

6 

7 

8 

9 

10 

11 

12 

13 

14 



and thus 

(7) provide an incentive for him to convert to System/3 60 

from non-compatible families. ( Id. , pp. 35-40.) 

Such benefits did, in fact, accrue to customers. For example, John 

Jones, Vice-President, Management Information Services at Southern 

Railway testified that compatibility was of "very great benefit" to 

him as a user because 

"it provides me the option of changing or upgrading the capa- 
cities and capabilities of my installed network and gives me 
an alternative which under some circumstances is a very 
attractive one in that I do not have to do reprogramming if I 
choose not to do it." (Tr. 80007-08; see also McCollister, 
Tr. 11068; Friedman, Tr. 50377; Case, Tr. 73427-28; Knaplund, 
Tr. 90507-08; PX 1215, p. I.) 



Since System/360 was compatible over a far broader range of processor 
capacities than any previous EDP line, those advantages of compati- 
bility were made available "to a great many users of all sizes" — from 

jj 

15 ! the large, multiple-location user who would be able to reduce his 

16 If training, system development and programming costs to the small first- 



17 ! 

18 | 

19 : 

20 ! 

i 

21 | 

22 | 

23 J 

24 J 
25 



time user who could plan to grow rapidly without incurring reprogram- 
ming costs. (Knaplund, Tr. 90507-08.) Of course, this meant that a 
"great many users" would be attracted to System/360. As Brooks 
testified: 

"We believed the compatibility would make it possible to 
make machines a lot easier to use, that it would serve the 
customers better, and that it would permit IBM to furnish a 
better level of customer support. ., . . [M] aking a machine more 
usable makes it more marketable." (Brooks, Tr. 22692; see 
also Case, Tr. 73427-28.) 

Joseph Rooney, who held a position as an IBM Branch Manager 

and later became the President of RCA's Data Processing Division, 



-298- 



1 ! 

2 j 

4- j 

5 i 

i 
i 
I 

6 j 

i 

7 j 
i 

8 ! 

9 ! 
10 j 
111 



testified that there was a "high degree of program compatibility" 

within System/360, which provided an advantage to IBM in that 

"Their clients could grow from a smaller system to a larger 
system, or if the economic situations were such that they wanted 
to go to a lower system,*- they could do so without having to 
reinvest in their software. It also was an advantage if you had 
a multi-faceted organization that had large computers and small 
computers, and some commonality of applications that they wanted 
to use on both types of systems. It gave the client the advan- 
tage of not having to modify his software to do so". (Rooney, 
Tr. 12550-51; see also Spangle, Tr. 5026; Beard, Tr. 10325.) 

Withington testified that " [t]here is an advantage to a 

manufacturer in standardizing on a single system set of programs 

because that minimizes his total cost of development, maintenance and 

customer support of such 'systems programs". (Tr. 56612.) In addition 



12 | to the tremendous competitive advantage* that IBM would derive from 

13 j of fering users a compatible family, the SPREAD Committee recognized 

14 ; that compatibility was "clearly advantageous to [IBM's] development 

15 | and manufacturing". (DX 1404A, p. 8 (App. A to JX 38).) Commonality 

i 

16 I in processor logic and programming were anticipated to provide IBM 

i| 

17 :|with economies in training of field personnel, development of program- 

18 



19 
20 
21 
22 



* The SPREAD Committee anticipated that this advantage was one 
that competitors would not be able to overcome during the rest of the 
decade unless they adopted new approaches to the achievement of 
compatibility: 



"Competitors appear to be relying heavily on common programming 
languages to achieve compatibility. The new processor family 
i| guarantees to IBM a compatibility level which will not be pos- 

23 ij sible, in the 1965-1970 period for a non-compatible family of 

! processors relying on common programming languages." (DX 14Q4A, 

24 ij p. 40 (App. A to JX 23).) 

25 ijAs we discuss later in the testimony concerning specific companies 
jj(see pp. 383-84, 480-32, 619, 623, 644, 660-61, 696, 705, below), a 

number of IBM's competitors did just that, albeit several years after 
IBM. 

-299- 



L; 

/^ 

3 ! 

4-1 



ming and standardization of installation and maintenance procedures. 
(DX 1404A, pp. 36-41 (App. A to JX 38); see also PX 1215, p. 2.) 
System/360 compatibility permitted IBM to realize these and other 
benefits. Case testified that training of programmers, salesmen and 
systems engineers was made "considerably easier" because they had to 
be trained for one group of machines instead of for different incom- 
patible machines* (Tr. 73387-88.) IBM also achieved cost reductions 
in manufacturing because of the ability to share parts among the 
various models of System/360 and to provide common training to manu- 
facturing personnel. . (Case, Tr. 73388.) Finally, IBM had to develop 
fewer operating systems than it would have for incompatible processors , 
and the design of the individual models was facilitated because 
commonality of design permitted the various engineering groups to 
T . ; l communicate effectively and assist in one another's design efforts. 
(Case, Tr. 73388-89.) 

The decision to provide a compatible line over a large 
performance range' was recognized within IBM as a risky one. From 



/ 
3 ; 

Q 

10 
U 
12; 
13 



IS 

17 | 
IS •! 



i 

a competitive standpoint, the SPREAD Committee anticipated that a 



| single compatible line could be marketed against by competitive 
! salesmen who would be able to develop "knock-off s" applicable to the 
i entire family. It would also provide a more nearly unitary target 
i against which competitors might react more effectively with their own 

22 :! 

.(product and price moves. (DX 1404A, p. 40 (App. A to JX 38).) Perhaps 
I most importantly, it would "encourage competition to be compatible 

** A \ 
£- I 

1 with [IBM] in order to tap [IBM's] support efforts". (DX 14 4A, 



£2 



-300- 



1 i 

2 i 

i 

3 ! 

4 j 

5 i 

i 
I 

6 | 

i 

7 ! 
8 
9 ! 

10 
11 
12 
13 ! 



p. 40 (App. A to JX 38).) That latter possibility was one that IBM 
plainly foresaw throughout the 1960s and one that came to fruition in 
different ways in the latter half of the 1960s and in the 1970s with 
the explosive growth of leasing companies and the advent of plug- 
compatible peripheral and CPU suppliers. (See pp. 750-96, 807-14, 

819-26, below.) 

The compatibility objective presented risks from a technical 
standpoint as well. Just as the attempt to combine business, scienti- 
fic and other applications in the same line raised the possibility 
that the new system would do none of them as well as a more specializejd 
machine, so too the attempt to achieve compatibility between very fast', 
processors and relatively slower ones raised the possibility that none 

!of them would be truly optimal. Case testified that 



14 ;| "It was thought prior to System/3 60 that having one machine 
ij architecture for both the fastest and the slowest machines in a 

15 j product line and, in fact, all places in between, could not be 
■j right because either the fast machines would be unnecessarily 

16 II restricted in the amount of function and capability that they 

jj could provide ... or alternatively, that the slowest and cheap- 

17 ;j est machines would be far too expensive by virtue of having to 

;j provide the richness of the instruction set that was provided by 

18 ! ! the larger and more expensive machines in the product line." 
ij (Tr. 73520.) 



19 
20 
21 
22 
23 
24 
25 



According to Evans, the "real challenge" of System/360 from an archi- 
tectural standpoint was to build a compatible family with a perfcrmanc 
range of 1 to 100 from the smallest machine in the family to the 
largest — it was "something that had never been done before".* (Evans, 



* The 3 60 announcement letter stated that the processors covered 
a performance range of 50 to 1. Evans called this a "conservative 
statement" and stated that the performance range was 100 to 1 at 
announcement and had since been expanded to nearly 1,000 to 1. (Tr. 
101177-78.) Evans testified that IBM successfully met its challenge 
and that System/360' s "performance range, unprecedented in the indus- 
trv", was a major factor in attracting customers to the 3 60. (Tr. 
101144.) 

-301- 



L, 

z 

/ 

3 

g 

10 

LI 
12 i 

14 



Tr. 101057-53.) The difficulty of this undertaking was clearly 

recognized by the SPREAD Committee: 

"It is not evident that downward compatibility can be attained 
through the whole product range. The group recommends, how- 
ever, that the design requirement for downward compatibility 
be stated as. a firm ground rule and that development proceed 
on this basis until the Phase I review. If, at that time, it 
appears that economically competitive downward compatibility 
cannot be achieved across the whole processor range, then the 
range shall be broken into two segments with downward com- 
patibility to be achieved within each segment." (DX 1404A, 
p. 17 (App. A to JX 38) .) 



17 ; 



Enfield, President of The Computer Software Company and 
former IBM Product Administrator for the DOS operating system, testi- 
fied that downward compatibility was achieved for System/360 through 
the Model 25. (Tr. 19977; see also Case, Tr. 73520-24.) For IBM to 
achieve that level of compatibility without incurring unacceptable 
;[ expense or performance penalties at the low end of the line required a 
"technological change in the way computer systems were built ... in 
*~ ,1 IBM" . (Case, Tr. 73520.) That technological change was the introduc- 
ia . tion of microprogramming or "firmware". ( Id . ) 

Microprogramming was invented by M. V. Wilkes of Cambridge 
-S :, University in 1951. (JX 33, o. 5.) Case testified that IBM was the 
i- '\ first computer manufacturer to use firmware in the building of computers J 

.J i 

£- ; j (Tr. 73222.)* Its use required the application by IBM of "new technical | 

2Z 'components" (such as transformer and capacitor read-only storage) and 

.i I 

22.1a new design "discipline". (Case, Tr. 73521.) ] 

„- 1 ' i 

-* f 

,'i i 

2± I * That use began with the experimental SCAMP built at Hursley in 

1 1960 (JX 33, p. 5) and continued with System/360. IBM continued its j 

25 ! innovations in "firmware" later in the decade with the invention of 

jthe floppy disk. (Case, Tr. 73223.) I 

I : 

! -302- i 



1 I 

2J 

3! 

I 
i 

I 

4 | 

5 j 

6 i 

i 

7 i 
i 

8 i 

s| 

I 
10 I 

I 

ii ! 

i 

12 | 

i 

13 ! 

i 

14 | 

15 i 

i 
i 

16 i 



Through the use of firmware (rather than hardware or soft- 
ware) IBM was able to achieve a number of the design trade-offs which 
System/360 required.* It was the "technical device . . . most 
responsible" for the fact that IBM System/360 computers were able to 
be designed efficiently for both business and scientific applications 
(Case, Tr. 73225; see also Evans, Tr. 101142-43) , as well as the 
method by which IBM was able to achieve full upward and downward 
compatibility. ** 

Some measure of the success that IBM achieved in imple- 
menting the architectural objectives laid down for System/3 60 may be 
gleaned from the longevity of that architecture. Compatibility and 
applicability to a wide range of applications were characteristics 
(assuming that they were effectively implemented) that would undoubt- 
edly be desirable in future systems. Accordingly, Case testified: 



"We tried to develop the computer architecture which would 
be extendable, which would be useful not only for the machines 
that were going to be announced in 1964, but also for subsequent 
l! machines as far into the future as we could plan for. ... We 
*' :! were thinking in terms of 15-20 years . . . and we would like to 
j have had that last even longer if that were possible." (Tr. 

18 j 73347.)/ 

19 '' 
20 
21 j 
22 
23 
24 
25 



* The need for such trade-offs was understood by the SPREAD Com- 
mittee, which imposed as an "engineering ground rule" the use of 
| microprogramming controls unless "conventional" control systems could 
jiprovide a cost/performance improvement of better than one-third. (DX 
|!l404A, p. 20 (App. A to JX 38).) Microprogramming was used in the 
i!system/360 processor models 2020, 2030, 2040, 2050, 2065, 2067 and 
"2085. It was not used in the 2044, 2075, 2091, 2095 and 2195. (JX 
38, <f 8, pp. 5, 6.) 

** As we explain later, it was also the means by which System/3 6 
enabled users to run programs written for earlier IBM computer systems , 



|| r To lengthen the life of the 360 architecture, the 360 design 
jjcrouD chose a memory addressina structure that provided for the eventual 

II -303- 



L i The architecture of System/360 lasted through the 370 into the 303X 

Z\ and 43XX lines and continues to the present time. (Evans, Tr. 101133; 

S || see also H. Brown, Tr. 82972; PX 4505, p. 1; PX 4531, p. 1; DX 860, 
4-i p. A; DX 9405, pp. 552, 1013.) 

5- ji (v) Emulation. Withington called IBM's introduction of 

r 
g- the System/360 "a substantial risk" for two main reasons: 



"One, IBM adopted a new machine architecture and a dependence on 
systems programs to cause the machine to be usable to the users. 
This was a large step in terms of the evolution of machine 
architecture and design, and it was not immediately certain 
either whether it would work well or whether the users would 
accept it. 

"The second primary area of risk was the lack of compati- 
bility between the 360s as announced and the predecessor IBM 
machines . 

"It was immediately obvious that the willingness of the 
customers to reprogram from the older machines to the 360s was a 
major question relating to its probable degree of success." 
(Tr. 56592-93.) 

The disadvantage of offering a new incompatible line was 

clearly recognized by the SPREAD Committee. It was, however, a 

disadvantage that had to be overcome rather than avoided if the 

Committee's concept for the new line was to be instituted. As the 



7 
3 

9 

10 ; 

ii 

iz 

13 

1+ 

15 
16 
17 

si 

;l attachment of 16 million bytes of main memory without modification and 

2G i about 2 billion bytes with only a "small" modification. That eliminated 
jone of the "major reasons" that previous architectures had been short- 

ZL ;] lived: the limitation on the amount of main memory that could be effec- 
tively used with those architectures. (Case, Tr. 73347-4 9.) The 8-bit 

22 [byte was another factor which gave System/360 architecture greater 

j longevity than previous systems. It permitted the use of 36 in appli- 

22 : cations that required character sets which made those applications 
[difficult to achieve on the 6-bit byte and 7-bit byte computers which 

2d '{preceded System/360. (Case, Tr. 73349-50.) 



-304- 



SPREAD Report noted, "Since [the new] processors must have capabili- 
ties not now present in any IBM processor product, the new family of 
products will not be compatible with our existing processors." (DX 
•J 1404A, p. 12 (App. A to JX 38)', emphasis in original.) 

The SPREAD Committee anticipated that the new capabilities 
provided by System/360 would induce many users to switch to System/360 
despite the need to convert their programming. Indeed, for many of 
these users, the very fact that they wanted to implement new functions 
rendered the entire question of conversion moot: 

"... While incompatibilities are a marketing disadvantage, 
it should be noted that systems reprogramming will, in many 
cases, be required, independent of the processor used. This will 
occur whenever the user wishes to obtain the benefits of any of 
the following: 

;j 

|{ "a. Random access rather than batch processing 

"b. The integration of communication facilities 

"c. The simultaneous operation of multiple processors 

"d. Multiprogramming to achieve efficient on-line operation." 

( Id. , p. 12; see also Currie, Tr. 15184-85; Withington, Tr. 
57683-84.) 

In "many cases", therefore, the reprogramming effort involved 
1 
j in switching to System/360 would be no more than a "natural outgrowth" 

i j of the systems improvements that the user wished to achieve--improve- 






► jments that would require a new programming effort whether or not that 
•i 

L .1 user switched to an incompatible processor. (DX 1404A, p. 12 (App. A 

> ito JX 38).) However, the Committee also recognized that " [sjome 

I. 'customers [would] be dissatisfied unless an alternative [was] provided 

■j 
• jto permit utilization of [their] prior machine investment". (Id. , 



-305- 



L 
2 
3 

4-1 

si 

I 

6- j 

T\ 
3 ; 
3 ; 

10 I 

11 l 

12 I 

i 

13 | 

14- : 



p. 39.) IBM provided customers with that alternative in the form of 
emulators . * 

Other manufacturers of computer systems also recognized the 
desirability of facilitating conversion and provided users a number of 
aids, such as simulators**' and translators, 7 ^ to ease the transition 
between incompatible systems.^ As late as August 1963, IBM was still 
working on software simulation as a means of providing System/3 60 
compatibility with prior systems. However, work on providing con- 
version through emulation had commenced within IBM prior to that time. 
(Hughes, Tr. 34047-48.) On August 1, 1963, D. H. Furth, Corporate 
Director of Programming, sent a memorandum to Evans expressing the 
view that it was "feasible" to use read only memory control (micro- 
programming) to achieve compatibility. He wrote: 



* An emulator is a combination of hardware and software that 
^* j permits one computer system to execute programs written for another 
'! system. (JX 1, p. 45.) 

;| ** A simulator performs the same function as an emulator, but is 

17 !| implemented entirely in software. (Goetz, Tr. 17654.) 
■» 

IS ;j /A translator is a computer program that takes as input the source 
:| programs of a particular computer and translates them as closely as 

IS ■} possible to an equal program in the same or a different language that 
■\ would run on the equipment to which conversion is desired. (King, Tr. 

20 1 14769-70.) 
i 

'i 

21 I • // GE offered a 1401 simulator which permitted programs written for 
] a 1401 to be run on its 400 line and a 7090 simulator which permitted 

22 :! programs written for the 7090 or 7094 to* be run on GE ' s 600 line. 

j (Weil, Tr. 7029-32.) RCA developed a simulator that allowed programs 

23 -written for IBM's 650 computer to run unchanged on the RCA 301. (DX 
; 561, p. 13.) Honeywell offered a LIBERATOR program which translated 

2- j IBM 1400 Series programs into programs usable on the Honeywell 200. 

■I (R. Bloch, Tr. 7578, 7588-39, 7605-06, 7886-39; Goetz, Tr. 17652-54, 
25 'j 13822-23; Enfield, Tr. 20052-54; DX 6661, p. 6.) 

! 

;! -306- 



L 
I 

I 

Li 



"Since such a hardware simulation would appear to be very 
economical from the customer's point of view and since it would 
eliminate some half dozen simulators from an already mountainous 
Programming Systems load, it would appear reasonable to pursue 
the realization of this feasibility as part of the overall NPL 
program." (DX 2872.) 



3 

!! 

T I 
3 J 

i 

0; 

* i 



By October that recommendation had been accepted, and Brooks wrote 
, that "We are hopeful that microprogrammed simulation can add substan- 
;| tially to the bag of tools for aiding conversion". (DX 2900.) 

During 1964, IBM announced microprogram-based compatibility 
features on System/360 for the 1401, 1410, 1440, 1460, 1620, 709, 
7010, 7040, 7044, 7070, 7074, 7080, 7090, 7094, and 7094 II proces- 
sors.* (JX 38, pp. 30, 289, 292, 334, 526; DX 14305.) Withington 
testified that System/360 was "the first major use of microprogramming 
for purposes of establishing backward compatibility."** (Tr. 56606.) 
He also testified that 



3 ' 

£• : "implementation of emulation using control store and micropro- 
grams, while it is more expensive [than software emulation], is 
regarded by users as preferable in most cases because it is so 
|! much faster". (Tr. 56371-72; see also DX 2900.) 

■6 :! 

•j The provision of emulators on Svstem/360 afforded users a 

| hardware alternative to conversion. (PX 449, p. 9.) It permitted 

3 

; them to transfer jobs to System/360 and to concentrate on new application 

.3 jj j 

:j areas without immediately having to convert their existing applications, j 

a; j i 

i 



!j * Case estimated IBM's cost of developing the 1401 compatibility 
£; feature on the Model 30 as $200 thousand and the cost of developing 

: jthe 7090 emulator on the Model 65 as one-half million dollars. (Tr. 
n ; 74557-62.) 

- i 

i± ■■} ** Withington defined "backward compatibility" as "the use of 

(emulation . . . for the purpose of allowing programs written for a 
jc jmanufacturer ' s prior computers with different instruction sets to be 
~* '.executed on the newer computers". (Tr. 56606.) 

>j 

:! -307- 



if 



(JX 38, p. 30.) 

Although programs run in emulation generally ran slower 
than they would have if rewritten to run in native mode on the new 
systems,* they could be run effectively enough to permit users to 
forego reprogramming if they chose to do so. (Beard, Tr. 9057-58, 
c ![ 9956-57, 10029-30, 10313-19; see also R. Bloch, Tr. 7608-09, 7614-15, 
7 ,j 7881-82; McCollister, Tr. 11287-89; Rooney, Tr. 11853, 12395-96.) 

Goetz testified that emulators were generally considered an "effective 
means of running programs from one computer system on another". 
(Tr. 17655, 13778.) 

3ecause 360 was incompatible with IBM's second generation 
equipment the conversion from IBM's second generation equipment to 
360 involved as large a task for users as would converting to 
.' another vendor's systems. (Beard, Tr. 9058-59, 9953-60, 10324-25; 
McCollister, Tr. 11069; Goetz, Tr. 18935-36; Enfield, Tr. 200-20-21.) 
Indeed, in some instances conversion to non-IBM equipment would have 



3 
9- 



Li 



12! 

n 

15 

is it 



;i been easier than conversion to 360. Weil testified that GE was 

17 3 

!{. initially "overjoyed" with the announcement of System/360 because GE 

IS ;[ 

had introduced a system "designed to displace" IBM's 7090s and 

is i 

r 7094s and believed that "it would be easier ... to convert from 
20 :i 

1 the 7090/7094 to the 600 series" than to 360. (Tr. 7060-61.) 



22 * 

I * Of course, such programs might very well run faster in emulation 
22 mode than they had in native mode on the equipment for which they were 

; written. For example, Enfield testified that a 360/30 operating in 
2± 1 emulation mode could execute 1401 programs 3 to 3 1/2 times faster 

•| than a 1401. (Tr. 20263.) 

25 \ 

! -308- 



Jones testified that Southern Railway ran benchmarks which showed 

that conversion from an IBM 7000 Series system to an IBM 360 was 

"about equal in difficulty" to conversion to an RCA or Burroughs 

machine, but not as easy as conversion to a Univac 1108. (Tr. 79042- 

43; see also Hart, Tr. 81936.) 

Nevertheless, IBM was successful in getting users to 

convert to System/360 from IBM second generation systems. (Withington, 

Tr. 57680-81.)* One reason for that success was, undoubtedly, the 

benefits that users were able to derive from System/360 's improved 

price/performance and new capabilities. As Withington agreed, 

"if [users] perceive it to be in their economic interest, 
[they] will absorb the cost of conversion for the future benefit 
that they expect to receive from [a] newly acquired computer 
system". (Tr. 57677; see also Hart, Tr. 80222-24.) 



Hart, head of the Computer Science Department of the General 
Motors Research Laboratories, testified that his department went from 
a 701 to a 704 to a 7090/94 to a System/360. (Tr. 81938-39.) Several 
;! years after these changes, Hart wrote "conversion costs must be taken 

* || into account when changing computers; however, in retrospect, the 

3 ! value of each of the above changes far exceeded the costs incurred" . 

* ■j (DX 3753 (Tr. 80193).) He explained that improvements in sheer computer 

2 ! speed, reduced computation costs, and the availability of "new kinds 

i 
I j of capabilities" were all reasons for changing computer systems. To 

i 



3 ; * It should be noted that IBM's success in getting users to convert 
| was not unique. According to Withington, between 1964 and 1970, some 

i. 1 90% of second generation equipment users converted to a non-compati- 
| ble computer system of either the same or a different manufacturer. 

s '! (Tr. 57677-83.) 



-309- 



Lj 

2 J 

i 

i 

3 ! 

3 

0+ 

7 | 

3 

9 

10 

II ! 



decide whether conversion is justified, "you take into account the 
costs of making the change, the benefits which are going to result 
from the change, [and] determine whether the benefits exceed the 
costs." (Tr. 80222-24.) 

A similar cost/benefit analysis was performed by NASA, 
circa 1965. NASA had just made a "large purchase" of second genera- 
tion machines to lower its operating costs, when a "new series of 
equipments"* became available with multiprogramming capabilities, I/O 
flexibility, memory sizes, program logic and the ability to use 
remote I/O devices that made it 



"possible to effect a consolidation of [NASA's] ADP resources 
. . . into a powerful central complex without compromising 
. availability, quality or power available to any user. At the 
-^ jj same time the cost per computation of these newer machines was 
T - 1 considerably lower than their old second generation equiva- 
^ :! lents". (DX 5440, pp. 2-3.) 

^; NASA decided to convert "at the earliest possible time". ( Id. ) 

NASA's analysis of the conversion difficulties was: 



U3 



IS ;{ "This conversion has created a considerable workload and has 

resulted in overlapping of older and newer equipments with its 
attendant increased rental costs during the conversion period." 
(DX 5440, p. 9.) 



: | It concluded, however, 



17 

IS 

IS] 

] "The benefits from the more complex software and the flexibility 

20 i of the new machines far outweigh any conversion cost we mav 
| incur." ( Id. ) 

21 1 

ij Despite the powerful incentives that users had to incor- 

jporate System/360 's new capabilities, it seems clear that 360 would 
22 



24 1 

.j * The new equipments included IBM 3 60s, Univac , CDC and GE computers. 

?c ! ! (DX 5440, p. 5. ) 

— -| 

i 

,i 
I 

•! -310- 



have been far less successful without emulators. Xerox's Competitive 
Reference manual noted the success of IBM's emulation approach to 
converting second generation users to 360 (PX 449, p. 9), and 
McCollister testified that it was a "very widespread practice" in the 
late 1960s for IBM users to choose the option of emulation on 360.. 
(Tr. 11287-38.) An IBM Corporate Programming Study based on a 

: November 1967 customer survey estimated that "more than half of the 
systems hours now being used by our Models 30, 40, 50 and 65 are being 
used in emulator mode". (PX 2161, p. 3.) 

(vi) System/ 3 60 Software. Prior to the advent of operating 
systems, each programmer had to write instructions that would schedule 
his tasks and control the various equipments he required for his 
particular jobs. As computer systems became faster and more complex, 
it became increasingly important to manage efficiently the resources 
they provided. Operating system software relieved programmers of the 

, need to incorporate scheduling instructions in each program they 

; | wrote and, -in effect, turned over the job of scheduling to the computer 
J '\ 

,1 itself. According to Dr. Perlis, operating systems enabled users to 

;i 

'} "take advantage" of a computer's total processing power, including 
9 \ 



i 

i i 

i 
M 
_ i 

i 

r i 

V 

I 

T 

m 



;l 

£ i 

■j 

2 1 

1 



its multiprogramming and multiprocessing capabilities . (Tr. 1848-49; 
see also Welke, Tr. 17113; Goetz, Tr. 17476-77; Enfield, Tr. 20737- . 
38; Case, Tr. 73443.) 

Given the complex "new market demands" and modes of use at 



! which System/360 was being aimed — i.e., "multi-terminal, on-line, 



IS 



I real-time, multiprogramming operation" (DX 1404A, pp. 7, 8, 9, 54 
I (App. A to JX 38)) — it was imperative that IBM automate as much as 



-311- 



L ; possible the system's resource management task. IBM embarked on the 

2 ; creation of a set of operating systems of varying complexity.* The 

3 I most complex of these, OS/360, was particularly ambitious. 
4- i OS/360 was designed to let customers "make the maximum 

'■. possible use of the relatively greater speed of the . . . System/360 
5* central processing units". (Case, Tr. 73438.) Since multiprogram- 
ming was anticipated to be a "normal" mode of use, facilities (such 
as an interruption mechanism) were to be included to make multi- 
programming "easier, straight forward and efficient". (Case, Tr. 
73438-39.) In addition, OS/360 was to contain facilities that would 
permit programmers to develop applications more efficiently, optimize 
the utilization of peripherals and simplify maintenance. (Case, Tr. 
; 73438-41.) 

Within IBM, it was recognized that "no one [had] ever 
undertaken a programming task of [OS/360's] magnitude". (PX 1092, p. 
4; PX 1900, p. 8.) Dr. Perils called OS/360 a "really major effort", 

id ;f 

;j one which "generalized every aspect of operating systems known at the 

* 'J 

' time and tried to in a sense build a system that would be all things 

' to all men". (Tr. 1887.) Mr. Welke, President of International 
IS !j 

'[Computer Programs, called OS/360 "a major programming effort" which 
2Q 1 

.ranked "along with . . . the creat undertakings of mankind". (Tr. 

21 j . 

■\ 17313; see also Rooney, Tr. 12576.) 

22 ;i 



7; 

3 !| 

9 
10. 
LI ! 
12 
12 
14 



j * To account for the varying degrees of speed and complexity of 
2i ] operation that users might desire, I3M provided with 360 a "spectrum 

[of operating systems . . . each of which offered a different memory/ 
25 j function trade off for the customer". (Brooks, Tr. 22759.) 



•i -312- 

•i 

r 
,1 



' .1 



So ambitious an undertaking entailed significant risk, and 
as we shall see, OS/360 was quite costly and difficult to perfect. 
Apart from the difficulty of constructing the operating system at 
If all, there was the additional .risk that users would reject the multi- 
' programming environment — an environment that was most often not used 
in earlier generation systems. (Case, Tr. 73526.) That would mean 
that IBM's investment in the hardware and software needed to permit 
multiprogramming would be reflected in System/360' s prices and would 
have accomplished little more than to make the systems less competi- 
tive. In addition, OS/360 1 s "extensive" resource management, data 
management, languages, aids to program development and error recovery 
techniques did not come "without a price". (Case, Tr. 73527-29.) 
The use of those capabilities would take up auxiliary storage space, 
main memory space and time on the CPU — an "operating system over- 
head" . (Case, Tr. 73529.) There was a significant risk that users 
would be unwilling to accept such "overhead" for the richness of 

: | function provided by OS/360. (Case, Tr. 73528-30.) 

i 

I OS/360 did, in fact, run into "difficulties in design, in 



H 

•(correctness [and] in completion".* (Perlis, Tr. 1887.) However, 

n i 

:| "when the system finally worked it had properties that were beyond I 

3 i ! 

; about any other operating system around". ( Id. ; see also Palevsky, ! 

M • ! 

.jTr. 3180; Rooney, Tr. 12576; Currie, Tr. 15186; Welke, Tr. 17308-12.) j 

2 :| I 

jit must be remembered that OS/3 60 was only one of five general program- I 



4. 1 



| * Many other computer systems suppliers experienced similar diffi- 
5 jculties in designing complex operating systems during the 1960s. 
i(See discussion below, pp. 364-66, 479, 502, 568-72.) 



-313- 



L 
2 

Z 

4- 

3" 

S A 



8 It 

a 

la 

ii 

12 

13 
14- 

15 
17 
IS 

-i 

it 

20 



i 

ft .! 



ming packages that IBM announced in 1964 for use with System/360. 

(Brooks, Tr. 22759; McCarter, Tr. 88388; JX 38, \\ 9, p. 6.) The 
others — Basic Programming Support (BPS) , Basic Operating System 

(BOS) , Disk Operating System (DOS) and Tape Operating System (TOS) 
were less complex sets of systems software. These operating systems 
"worked reasonably well from the start" and were well accepted by 
j ;j customers. (Withington, Tr. 58596-600; Enfield, Tr. 20947-52, 21120; 
Brooks, Tr. 22853-54, 22862-63; McCarter, Tr. 88388; DX 1410; PX 6217, 

pp. 3-4.) DOS in particular, which was less complex than OS/360 but 
still 25 to 50 times as complex as the systems software provided with 
the 1401, was highly rated by users and widely used. (Enfield, Tr. 
20299-300, 20741-42, 20088-89, 20943-48.) Case testified that "if it 
had not been for the operating systems for System/360 . . . the value 
of that equipment to users would have been considerably less than it 
was and . . . the orders and acceptance for that equipment would have 
been a lot less than they otherwise were". (Tr. 73443-44.) 

(vii) System/360 Peripherals. Case testified that one of 
the design objectives for System/3 60 was to provide "a wide variety 
of peripheral equipment that could be combined in a very wide range I 

of configurations". (Tr. 73416.) Prior to announcement, the "breadth" j 

j 
!of 360 's peripherals were viewed within IBM as a prime motivation for ! 

I . | 

:j users to re-systemize their applications and convert to 360. Thus, I 

tin January 1964, Brooks wrote: ' j 

2j i ! 

| "Even though present applications can be simply mapped onto 

2.~\ System/360, many new system concepts will offer substantial j 

j incentive for the customer to re-plan his application. These 

2s 1 include file orientation, communication facilities, large memories, \ 

~~* bulk stores, etc." (DX 1172, p. 1.) j 

.1 : 

I ; 

•j -314- I 



The April 7, 1964, 360 announcement contained "many features 
different from those previously offered by IBM" . Included in the 
announcement were "direct access storage devices (including the 2311 
disk drive, the 2321 data cell and the 2301 drum storage device); 
; control units, high performance tape drives (including the 2400 series 
s ! and the 7340 Hypertape drive Model 3) ; visual display units (includ- 
ing the 2250) ; 7770/7772 audio response units; communication and data 
acquisition equipment (including the 1070 process communication 
system); and a printer, the 1403-N1". (JX 38, \ 6, p. 4.) IBM also 
announced numerous additional peripheral devices for use with System/ 
360 subsequent to the April 7 announcement — including the 2314 disk 
drive, new terminals, additional models of the 2400 tape drive, the 
2420 tape drives and optical character recognition equipment. (Id. ) 



3 i 
9- 

o ; 

x 

2, 

i: 



The 360 announcement letters describe some of these peri- 
pherals as follows: 

|j 1015 Inquiry Display Terminal: "Used to interrogate and 

£ if 

j| receive visually displayed replies from a System/360, mdl 30, 40 or 



.3 



|| 50." (JX 38, p. 43.) 

\ 1070 Process Communication Svstem: "A Tele-processing System 

L9 i[ | 

:} designed for two-way data communication between remote process loca- i 

zo:j j 

:[ tions and a central data processing area." Applications include "control i 

21 1 

Ij of oil fields, petroleum and natural gas pipelines, utility' distri- j 

: bution systems; data collection in refineries, chemical plants, steel ! 

a | ! 

[mills, and manufacturing processes .... The 1C70 forms a complete ! 

2A i ■ 

. tele-processing system when attached to . . . System/360, via a 2701 ; 



-315- 



Data Adapter or 2702 Transmission Control". ( Id. , p. 39.) 

1403 Model Nl Printer (originally announced as 2201 Mod. 3) : 
"[plrinted output for a System/360, model 30, 40 and 50. . . . maximum 
speed, 1,100 1pm". (IcL/- pp. 84, 198.) 

1418 Optical Character Reader; "Optically reads data from 
printer card or paper documents. ..." ( Id. , p. 70.) 

1419 Magnetic Character Reader: "Reads magnetically 
inscribed data from card and paper documents. . . . Documents read at 
maximum rate of 1600 documents a minute." ( Id. , p. .71.) 

2250 Display Unit: "A cathode ray tube unit for displaying 

output in alphameric and graphic form for System/360 An 

input/output unit which offers increased speed and flexibility for 
; file inquiry, inventory control and dynamic monitoring of computer 
operations and continuous process control." ( Id. , p. 85.) 

2301 Drum Storage: "High performance random access storage 
for a System/360, mdl 50, 60, 62 or 70. . . . [D]esigned for 
applications such- as main memory extension, programming system resi- 
dence and table or index storage." (JX 38, p. 86.) 

2311 Disk Storage Drive: "For fast, flexible access . . . 
85 millisecond average access speed . . . 156 KC/312 KD data rate . . . 



adapters enable the 2701 to communicate . . . with the following 
terminals : 

"1060 Data Communication System .... 

"1050 Data Communications System .... 

"1033 Printer ..... 

"1031 Input Station .... 

"1070 Process Communication System .... 

"1053 Printer .... 

"AT&T 83B2 Selective Calling Terminals .... 

"Western Union Plan 115A Outstations .... 

"Common Carrier TWX Stations .... 

"1009 Data Transmission Units, 1013 Card Transmission 
' j" Terminals, 7702 Magnetic Tape Transmission Terminals or 7740 
Communication Control Systems .... 

"7701 Magnetic Tape Transmission Terminals or 7750 
Programmed Transmission Control Units .... 

"7710 Data Communication Units, 7711 Data Communication 
Units, or another System/360. ..." ( Id. , p. 90.) 

2702 Transmission Control: "For on-line attachment of 



i 



various asynchronous input/output devices via private or commercial 
j common carrier transmission facilities to a System/360. . . . [A] 



;l modular unit with a variety of features to meet a customer's data 

f .! 
■» 

i communication needs with a System/360". (Id. , p. 9 3.) 



Multiplexor Channel: " [P]ermits simultaneous operation of 
I/O units on time-sharing principle . . . primarily designed to 



> 



[handle multiple terminals and low speed I/O units." ( Id. , p. 31.) 
L i 

J The combination of those and other peripheral product 



-317- 



1 i 

i 

2 i 

3 j 

5 I 

« ; 

7 | 

i 
3 | 

a ! 
LI I 

12. i 

i 
13 j 

is i! 



announcements and the announcement of six central processing units 
with a wide range of memory options was "unprecedented in the 
industry". (Evans, Tr. 101134; JX 38, pp. 14-25; see also PX 
4829, pp. 16-18.)* This range of peripherals was important to 
customers when considering System/360 against competitive systems 
because it greatly expanded their ability to change or add to their 
systems as their requirements changed and "played a large part" in 
customer decisions to go to 360.** (Evans, Tr. 101134; see also PX 
4829, p. 13; Withington, Tr. 56770-71.) 

The broad range of peripherals announced with 360 promoted 
two of the SPREAD Committee's primary objectives — the creation of a 
single system able to perform all applications and one that would 
address increasingly important new applications (i.e., multi-terminal, 
on-line, real time applications). The announcement of new disk drives, 
tape drives, communication controllers, card and printer I/O, ter- 



«i! 



1 * " [T]here has never been a time when any of the general purpose 

^' <l • competitors to IBM have offered more variations on peripheral equip- 
ment, the total breadth of applications and systems program functions 
and the total number of alternative processors" being offered by 

j IBM. (Withington, Tr. 56770.) 

19 i ■ 

; | ** We do not mean to imply here that all of the peripherals announced i 

-S .j with 360 were successful. A number were soundly rejected by users. j 

1 For example, the 2321 data cell was a "major product failure" which j 

2i |! failed to achieve success because of unreliability; IBM had to super- j 

j sede the 1015 terminal with the improved 2260 because the 1015 was j 

22 j simply not competitive; and Hypertape turned out to be a "failure" even j 
] though it was judged within IBM to be technically superior to com- j 

23 [ petitive offerings. (Case, Tr. 74205-06, 72787-88; Withington, Tr. j 
1 58534, 56475-76; JX 33, pp. 346-47; PX 6671, pp. 15, 26; PX 2990, p. R3 ti 

24. | DX 13949.) As we discuss below, IBM acted quickly to shore up areas j 

;j in the product line which were not judged to be superior to competitive . 

25 j offerings. (See below, pp. 390-95.) ! 

I ! 

t ; 

j -318- \ 



L j minals, audio response equipment, magnetic and optical character 

2. : | readers and paper tape and process control units meant- that users 

3. : could build configurations specifically tailored to their application 

i - 
4.|- requirements — whatever those requirements. Dr. Gibson testified that 

; one of the features of 360 that permitted it to be used for both 

scientific and business applications and "erase the previous distinc- 

j \ tion" was "the very wide range of input/output equipment easily 



3 

9 



i 

Z 



attachable through a common interface, . . . [which] made it relatively 
simple to configure a commercial system ... or one. optimized for 
scientific computing". (Tr* 2948-49; see also JX 38, p. 28; PX 3638, 
p. 1; PX 4829, p. 13.) 

In addition, the variety of remote I/O and communications 
equipment offered with System/360 underscored 360 's emphasis on new 
applications. Weil of GE wrote that System/360 "has major strength 
in a variety of new mass storage devices and a whole new array of 



; remote terminal equipment .... It has many of the features which will 
: [ make possible its application in direct access systems." (PX 320, p. 

■n 

'(13.) Displays, remote data collection equipment, remote process 
'{control equipment, communications controllers, data communications 

' equipment and on-line banking equipment were all made available to 

20 !i 

'{permit users to bring the power of 360 to bear at the point of trans-. 
11 ] 

.j action — in real time. The ability of a System/360 to communicate 

22 ! 

[with other computers or terminals "opened ua a whole new gamut of 

a i ' 

j applications in industries, airline reservations industries, modern 

i- ; 

■j business, so that remote stations could have access to the enormous 

?s i 

1 -319- 



L 
2 
3 

5 






data in a central computer and do so in real time". (Evans, Tr. 
101136.) The ability to do such applications resulted in sales of 
systems that otherwise would not have been sold. (Evans, Tr. 101135.) 

The importance of System/360' s peripherals to the success 
.of the product line cannot be overestimated. As Mr. Norris of CDC 
testified, the speed, performance and price of peripherals are "impor- 
tant considerations in determining to acquire one system or another" .. 
(Tr. 6019-20; see Withington, Tr. 56239, 56246-47.) Thus, even a 
single peripheral device— such as a disk drive, terminal or printer — 
which is sufficiently better than competitive offerings can swing the 
total system decision. ( Id . ; Currie, Tr. 15495-96; Rooney, Tr. 
12048-49; DX 13949.) In this respect, of all the peripherals offered 
with System/360, the 1403 Nl printer and the 2311 and 2314 disk 
drives were most critical to 360' s success. 

1403 Nl Printer. We discussed earlier the importance of 
the 1403 printer to the success of IBM's 1401 computer system, and 
how that printer gave IBM a "tremendous advantage" in the marketing 
of systems until competitors began to offer "satisfactory alternatives" 



7 
3 

ia ; 

u 

12. 
13 
IA 
15 

*6 , 

IS ; 

ijby 1963 or 1964. (See above, p. 143.) In 1964, IBM announced 

13 

:|the 1403 Nl Printer for use with System/360. The 1403 Nl ran at 

20 J. 

[almost twice the speed of its predecessor (1100 lines per minute 

21 1 . 

:j compared to 600 for the 1403) and cost only about 15% more than the 

22 l \ 

[1403. (Evans, Tr. 101137; DX 3617; see also Enfield, Tr. 20266; JX 

j38, p. 207; DX 573, pp. 4, 6.) At the time of its introduction IBM's 

<■■* i 

.{competitors did not offer a printer that matched the 1403 Nl in print 

25 i 

J ' -320- 



quality, price and speed. (Evans, Tr. 101137; see also Case, Tr. 
72881.) IBM's competitors recognized and acknowledged the excellence 
of IBM's printers . Beard (former Chief Engineer of RCA's Computer 

| System Division) testified that RCA began offering the 1403 with its 
Spectra Series because there were applications for which customers 
desired print quality "of a very high standard" . Such customers 
"insisted" on "1403 chain printer type quality" and "after resisting 
these requests some period of time" RCA acquiesced and "put the 1403 
into the RCA computer line". (Tr. 10322-23.) 

The 1403 Nl was particularly important to System/360 's 
ability to perform certain business applications. For a customer with 
applications such as payroll, billing, accounts receivable and inven- 
tory control, the ability of a computer system to do his work is 
determined "in large measure" by the speed, quality and reliability 
of the printer. (Evans, Tr. 101137; see also Currie, Tr. 14971-72; 

; Withington, Tr. 56253.) 

Currie testified that XDS was at a "disadvantage" to IBM 
with respect to its line printer for customers that wanted to do "any 

.! significant amount of business data processing". (Tr. 15459.) As 

'j late as 1969, XDS was only "marginally competitive" in peripherals and 

► 'I 8 

i i j 

1 its line printers "were not acceptable to some of our users". Those . I 

j printers lacked the range of "speed/performance" that some customers | 

I 1 I 

| wanted and did not produce as high a quality print as a chain printer j 

i t i 

lor a train printer. (Currie, Tr. 15006-08.) ! 

' ! 

;! CDC also experienced "substantial problems" in marketing some 



-321- 



t 

2 

3. 

4- 

S 

T\ 

a | 

si 
i 

i 
u! 

12 



1* i 

i 

i 



of its computer systems because they incorporated printers that "lacked 
sufficient reliability to meet normal customer expectations" and had 
"a poor print quality, in terms of wavy print". To help solve these 
problems CDC acquired the Printer Division of Holly Carburetor in 1966. 
(G. Brown, Tr. 51528-29.) CDC ultimately developed a "1403N-1 type"" 
printer of its own, but it had to be "reworked and re-developed" in. 
the 196 9-70 time frame in order to effect reliability improvements. 
The changes resulted in a design that was "more like the original IBM 
design". (G. Brown, Tr. 51541-46.) 

While CDC attempted to copy the 140 3 Nl design and RCA simply 
incorporated it into RCA's product line, Grumman Data Systems took 
advantage of the 1403 Nl's superiority by offering to attach it to a 
number of non-IBM computer systems. As late as 1975, an advertisement 
for Grumman Data systems stated: 



ia 



"For years people have been trying to imitate the IBM 
1403. Unsuccessfully. Now, with the Grumman Printer Controller 
you can connect your present computer to an IBM 1403 and give 
yourself the best printing in the business. 



17 ;[ "The IBM 1403 has built an extraordinary record. Highly 

|| reliable, high speed operation. Unusually consistent, clearly 

IS jl readable printouts. (No wavy lines so typical of drum printers.) 

i! Type fonts your operator can readily interchange. And, of course, 

19 ;i it handles form changes easily. 



25 JGrumman later offered the 1403 Nl for attachment to Burroughs, Data 



20 I "With the Grumman Printer Controller you can improve your i 
| printing quality, speed, and reliability. All at an attractive,. j 

21 -| and perhaps, money-saving price. Speaking of price, you can buy i 
• our controller or rent it. We provide maintenance of course. ! 

22 [ t . ! 
"With our printer controller you can connect the IBM 1403 | 

to your present DEC, Xerox, GE , or CDC computer. We'd like to j 

I hear from Burroughs, Univac and the other computer users, too." ! 

24 , | (DX 94B.) j 



-322- 



General, Digital Scientific and Univac computers. (DX 2782A; DX 7984.) 
The 140 3 Nl was also offered with Computer Machinery Corporation 
computers. (DX 11665.) 

Gordon Brown testified that the quality and reliability of a 
■. printer is "an extremely important criterion in the selection of a 
computer system". (Tr. 51528-29.) The 1403 Nl was a real boon to 
the acceptance of 360. 

System/360 Disk Drives. As we discussed earlier, IBM's 
superiority in direct access storage technology during the 1950s and 
early 1960s contributed greatly to the success of IBM's first and 
second generation systems. (See above, pp. 91-95,149-53.) IBM maintained 
that superiority with the disk drives introduced for use with System/ 
360. Both the 2311 and 2314 were substantial improvements over IBM's 
earlier disk drives and both proved critically important to the 
success of System/360. These disk drives were more than just superior 
to competitive offerings, they were unique in the industry: there 
simply were no similar competitive offerings for several years after 
their introduction. Thus, they gave IBM a competitive advantage in 
the marketing of 360 systems that competitors were unable to match 

until the late 1960s, and even then, competitors were able to do so 
I J 

I only by adopting, in one way or another, IBM's disk technology. 

:! 

■\ IBM announced the Model 2311 disk drive on April 7, 196 4. 

I ■;; 

;| The 2311 had approximately twice the access speed, twice the data 

• i 

irate and two and one-half times the storage capacity of the 1311. 
i. 1 

|(Case, Tr. 72741-42; JX 38, p. 86; PX 4252, p. 1; DX 3554D; see also 



I -323- 

•i 



I 
z 

3 

4. 



3 

e 

71 
3 
9- 
IQ 

IZ! 

Is ! 

14 

15 

IS ;i 

17 J 

IS 

: earlier model". (Tr. 9597.) Withington agreed that both the 2311 

19 i - 

itand 2314 were unmatched by comparable competitive products during the 

2d ! 

{initial years in which they were marketed. (Tr. 58800, 56240-41.) 

21 i 

:| IBM foresaw and depended upon the widespread acceptance of 

22 :, » 

idisk. drives as a key factor in the ultimate success of System/360. 
>•«» .' 
cz ; 

I IBM Vice President Knaplund testified: 

24 I 

j "An important element of the System 3 60 forecast was the antici- 

ys. ' pation that. disk files would be used extensively, both in applica- 

-324- 



Enfield, Tr. 20264-65; Haughton, Tr. 94998.) 

IBM announced the 2314 disk drive on April 22, 1965. The 
2314 had a faster access speed, double the data rate and almost four 
times the storage capacity per spindle of the 2311. (Case, Tr. 
.72742-43; JX 38, pp. 86, 439; DX 3554D; see also Haughton, Tr. 94998.) 

Beard testified that the 2311 represented a "technological 
advance" over prior random access storage methods. "It provided not 
only . . . fast access time but it provided . . . for the first time, the 
degree of reliability that was required of random access devices 
.... [I]t was really the first very reliable disk file that . . . 
was offered by anyone". (Tr. 9048-49.) Beard also called the 2314 
an "advance over prior random access devices", adding that his comments 
on the 2311 applied "perhaps more importantly" to the 2314 because 
the 2314 offered greater storage capacity and a more "practical cost" 
for random access storage than did the 2311. McCollister testified 
that the 2314 was "[v]ery definitely" an advance over prior disk 
drives because, for example, "it had a capacity in a pack of approxi- 
mately 28 million bytes as compared with 7 1/4 million bytes in an 



y ;l 



\ 



tions that had historically utilized magnetic tape or punched 
card storage and in the development of new communications 
oriented — or 'teleprocessing' — applications." (Tr. 90506.) 

However, the demand for the 2314 disk drive "" turned out to be very 
J' surprising in the rate that customers found use for it". (Case, Tr. 
72743.) IBM "totally underestimated the demand for such devices" and 
"we [in IBM] found ourselves hard pressed to deliver the devices as 
fast as customers were demanding them" . ( Id. ) It is important to 
note that the use of disk drives was not common on second generation 
computing systems. According to Case, fewer than twenty percent of 
computer systems prior to 1964 used direct access storage devices. 
(Tr. 73527.) Nevertheless, IBM "gambled" that System/360 would be 
widely used in "operational- type" applications (as opposed to batch- 
type applications) and that disks would play a "pivotal role" in such 
applications. (Evans, Tr. 101139.) System/360 's more advanced 
operating systems, were designed in a way that required a direct 
access storage device for their successful operation. The higher 

jj performance and greater function necessary to achieve such operation 

It 

:| could not have been provided with magnetic tapes and the use of drums 

j would simply have made the cost of storage too expensive. (Case, Tr. 

i 

j 73451-53.) IBM was therefore betting that users would be willing to 

! 

I 

j trade-off the expense of disk drives for the increased efficiency of 
i 

! 

j operation and the additional function that a disk-based system would 
i 

; 

be able to provide* — that users would accept widely an approach to 



* The "significance" of the disk drive was that it provided a 

[functional capability of having information on-line and readily 

javailable. (Rooney, Tr. 12142.) The random access capability of 

t 

i 

I -325- 



computing that had not been widely accepted before. 

In hindsight, that bet was a good one. As Case testified, 

today "nobody thinks of developing a wide range of computing equipment 

or a family of computer systems without having a direct access storage 

device as a prerequisite for the operating systems". (Case, Tr. 

73452-53.) Back in 1964, however, nobody but IBM had that thought or 

acted upon it as forcefully.* As a consequence, the tremendous 

acceptance of IBM's disk drives swept before it all of the other 

approaches to random access storage then being offered: 

"During that period the entire industry and the users began 
to appreciate the importance that disk drives were going to play 
in the great majority of general purpose computer systems. 
Before that time, alternatives were being experimented with, such 
as particularly magnetic card devices, and also I think no one 
realized the degree to which the transaction processing mode of 
use was going to prove popular. I believe only IBM among the 
major competitors at the time offered an alternative between 
magnetic card devices and disk drives, with developments pro- 
ceeding along both lines. A number of the other manufacturers 
committed themselves almost entirely to the magnetic card devices, 
sometimes also using magnetic drums. 



L 
2 
Z 

4* 
5 

7 ! 
3 

? 

lq 

as 
iz 

13 

i t 

IS 

17 

IS 

ta 

23 ;■ disks "permitted a new and more effective approach to doing customers' 
;jwork", particularly in real-time applications such as those performed 

21 -| by banks and airlines. (McCollister, Tr. 9591.) System/360's empha- 
ij sis on disk drives made possible more efficient use of CPU, main 

22 : ; memory and peripherals; increased the range of functions and services 
jthat could be provided by the operating system; and made possible a 

23 {"more valuable" mode of operation (random processing of transactions) 
[than the sequential access mode of processing that was common prior 

24, |to the emphasis on disk drives. (Case, Tr. 73463-70.) 

-c j * As we discuss below NCR, Burroughs, Sperry Rand, Honeywell and 

jRCA all offered different approaches to random access storage, and 

ijall of those approaches failed in the face of the tremendous user 

(acceptance of disk drives. (See below, pp. 94 383, 473-74, 549-50 

; !659.) 
I 

;i -326- 



"When it became apparent that the class of magnetic card 
devices was not going to be successful in the marketplace, for 
reasons of reliability, and that the disk drive was a critical 
product, many of IBM's competitors were left for a while without 
a satisfactory option." (Withington, Tr. 56240-41.) 



n 

_ i 

3 

3 

a 



Both the level of performance and the attractiveness of 
System/360 were substantially dependent on the 2311 and 2314 disk 
drives. (McCollister , Tr. 9370, 9591-92; Rooney Tr. 12122; Knaplund, 
Tr. 90506-07; Evans, Tr. 101138.) The 2.311 was "far more" important 
ito the marketing of System/360 than the 1311 had been for IBM's earlier 
systems, because the 2311 "offered an improved price/performance . . . 
was supported to a greater degree by systems programs . . . and, 
therefore, was easier to use, and . . . was more reliable". (Wi thing- 
ton, Tr. 56246-49.) And the 2314 was, if anything, even more impor- 
tant. It provided "a functional capability very much needed in terms 
of price/performance in the competitive marketplace and without that 
capability you were in a weak competitive situation against IBM" . 
„ ;[ (Rooney, Tr. 12193.) Within IBM the 2314 was recognized as a "catalyst 



4- 
5 
6 



to make many systems sales for previously undeveloped application use 
of computers" and as a "door opener that beats competition". (PX 1967, 
pp. 1, 3, see Page, Tr. 33122.) According to Case, IBM's emphasis on the: 
use of disk drives with System/3 60 contributed to the objective of 
growing the market for IBM products in particular and computer system 



ij products in general. (Tr. 73468-70.) 

Q :! 

'1 Not surprisingly, other systems suppliers wanted the kind 

rr 'J 

■|of "catalyst" for systems sales that IBM already had. Eventually, they 
[I | 

j either acquired them from OEMs or from IBM itself or they undertook to 

2. :i 

; manufacture them themselves. As we discuss below, the acceptance of 
13 ■ 

•360 spurred the growth of peripheral equipment manufacturers, some of 

jwhom supplied IBM 2311 and 2314 type disk drives directly to IBM end 
— .1 

I 

! -327- 



Lj users. During the latter part of the 1960s, however, these manufac- 

2 : turers served as a prime source of disk drives for many systems 

3; suppliers. (See pp. 753-59, below.) 

4- i- Memorex was the first of the PCMs to offer IBM plug-compatible 

3 if disk drives, in 1968. (See p. 770, below.) During the years 

1967-70, Memorex hired almost 600 former IBM employees, three of whom 
became Memorex Vice-Presidents. (JX 34, pp. 1-2.) In 1967, Memorex hired, 
a number of disk drive engineers from IBM, including Roy Applequist, 
who had designed IBM's voice coil actuator. (Guzy, Tr. 32858-64; 
Gardner, Tr. 38585, 39143*) Applequist designed the voice coil 
actuator for Memorex 's 630 disk drive, which, according to an indepen- 
dent engineering assessment, was "directly derived" from IBM's 2314B 
(3330) and "not the result of coincidence". (Gardner, Tr. 39143; DX 
1418, p. 151; see also Spitters, Tr. 55259-61; DX 2572.) D. J. Guzy, 
former Executive Vice President of Memorex, testified that the hiring 
^g -{of Applequist and other IBM engineers was important to the success 

£7 ; ithat Memorex achieved with the 630; and that the 630 and 660* were 

'I 
IS , styled and intended to be, respectively, 2311-type and 2314-type disk 

19 j drives. (Tr. 32316, 32776, 32899.) Memorex marketed the 630 and 660 

2G i not only directly to IBM end users, but also to a number of different j 

■' i 

21 J systems manufacturers, including RCA, Univac, DEC, Burroughs, Honeywell,; 

■i i 

22">SEL, Hewlett-Packard, Siemens, Phillins and ICL. (Guzy, Tr. 33168; j 

23 jDX 1302, pp. 1-3; DX 1308, p. 1.) 

24 1 

:i 

2z I * Memorex did not begin volume production of the 6 60 until the 
i second quarter of 1969. (DX 1268, 0. 17.) 

:i 

•! -328- 



7 

3 

9- 

10 

II ! 
12! 
13 • 
14 



1 I 

2 I 

3 ! 
4 
5 
6 
7 
8 
9 

10 

11 

j 

12 | 

i 

13 i 



15 
16 



ISS was formed in December 1967 by twelve former IBM employ- 
ees who had resigned from the San Jose Laboratory, where they were 
responsible for disk drive development. A number of this so-called 
"dirty dozen" had worked on IBM's Merlin (3330) program. (Whitcomb, 
Tr. 34555-56; DX 4756B, p. 96; DX 4739: Wilmer, Tr. (Telex) 4266; DX 
4741: Yang, Tr. (Telex) 6116.) Like Memorex, ISS manufactured 2311- 
type and 2314-type disk drives, the 701 and 714, which were marketed 
by Telex to IBM end users beginning in 1969. (PX 4732A, p. 12; DX 
4242, p. 8; DX 4250, p. 7; DX 4756A, pp. 36, 7 2.) ISS also marketed 
disks OEM to Hewlett-Packard, Itel and Storage Technology Corporation. 
(DX 86A, p. 2; DX 4113: Terry, Tr* (Telex) 3310-12.) The ISS 2311- 
type drive was similar to IBM's 2311 except for the addition of a 
voice coil actuator, and the ISS 2314-type drive was functionally 



14 j equivalent to IBM's 2314, again except for the addition of a voice 



coil actuator. (Page, Tr. 33072-73; Ashbridge, Tr. 34812-13.) ISS 
was eventually acquired by Sperry Rand (in 1973) for its advanced 



17 ;|disk technology, technical capabilities, highly qualified personnel, 



18 j 

19 | 

20 j 
i 

21 j 



plant facilities and highly profitable OEM customer base. (DX 86A, 
pp. 1, 4, 5; DX 87, p. 12.) After the acquisition, ISS became the 
developer and manufacturer of disk subsystems for use in Univac 
systems, but continued marketing 2314-type disk drives to IBM users 



22 ! iand to OEM customers. (Eckert, Tr. 988-89; McDonald, Tr. 4060-63.) 



23 i 

!■ 

24 I 
i 

25 i 



CalComp also offered 2311-type and 2314-type disk drives, 
manufactured by Century Data Systems, to end users and on an OEM 
basis. (DX 10735, pp. 10-11; see pp. 776-777, below.) CalComp 
shipped its first plug-compatible (2311-type) disk drive in June 1969 



-329- 



l'; (PX 5324, p. 46; DX 4756A, p. 8), and later became the "first company 
2'i to produce and ship a 2314 equivalent". (PX 3707A, p. 38; DX 10735, 

2 T p. 10.) Century Data marketed these disk drives to leasing companies 

i 
^ |. such as Randolph and to other systems suppliers such as Nixdorf, 

!h Burroughs and Univac (PX 3146A, p. 1; PX 5581, p. 10; PX 5582, p. 7; 

![ DX 1886, p. 7; DX 12194.) 

Similarly CDC manufactured and marketed 2311- and 2314-type 
disk drives, both end-user and OEM. CDC's OEM customers included 
Honeywell, GE, Siemens, RCA, XDS, ICL, SAAB, CI I, Burroughs and 
Telex. (G. Brown, Tr. 51056-57, 51080-81, 51095-96; see pp. 682-84, 
1074-77, below.) 

RCA did not even wait for PCM's to copy IBM's technology, but 
went directly to the source. "It was apparent [to RCA] that this 
capability which was offered by IBM was going to be required by RCA 
in order to successfully market its products." 



/ 
3 

Si 
10 

u 

12 

13 
14- 

"This capability at the time was not available from any other 

IS ;1 source. So, therefore, when we announced the Spectra 70 family 

it or series, which came out about eight months after the IBM 360 

17 i announcement, we announced as a part of the RCA product line this 

' particular Model 2311 disk pack file capability and we obtained 

15 ! these files by buying them from IBM, the same as any other 

;j customer would buy them from IBM." (McCollister , Tr. 9370.) 

19 : 



Although RCA had its own disk drive development program, RCA 

subsequently contracted with Memorex to supply disk drives for use with | 

! 

RCA computer systems because Memorex' development program was further 
ahead than RCA's "which was going to be about a year later than ! 
Memorex' s". (Beard, Tr. 8575.) RCA went to Memorex at a time "when ! 

i 

we had in parallel our own development going on" because RCA was "under • 



-330- 



parable product to the IBM 2314 at the time". RCA "couldn't afford 
in the marketplace to wait that additional year" necessary for RCA's 
development program to produce the required disk drives " [bjecause 
we were losing too many sales for the lack of it" to IBM. ( Id. ) 

GE, on the other hand, attempted to build an IBM plug- 
compatible 2 3 11- type drive. (Ashbridge, Tr. 34 812-13; G. Brown, Tr. 
51536-37; Spain, Tr. 90227.) But "it met with limited success and 
arrived to the marketplace much too late to meet market, or customer 
requirements". (G. Brown, Tr. 51536.) GE entered into an exclusive 
contract with Greyhound Computer Corporation to sell the device, but 
Greyhound ended up having to take a significant write-off on its 
investment in the GE equipment and even sued GE. (Spain, Tr. 88753, 
88755.) 

Not until the very end of the 1960s had IBM's disk tech- 
nology been sufficiently spread around the industry for some of 
IBM's systems competitors to have pulled even. Thus, the January 5, 
1970 Phase III Level Forecast Assumptions for IBM's soon-to-be 
announced Merlin* disk drive reported: 

'j "System Manufacturers 

j [ 

;| "From the announcement of the 2314 in 196 5 until late in 196 8 I 

'! IBM had significant competitive advantages in this product area, j 

■j as no competitor could offer a direct access device with the j 

\ price, capacity, performance, and interchangeability character- j 

j istics of the IBM 2314. The situation today, however, has changed j 

; ; radically as most system manufacturers now have announced devices j 

j which are virtually identical in specifications to the IBM 2314. j 

; : The chart below tabulates the status of the ten major system j 



* As we shall see, the Merlin (3330) drive put I3M right back in 
the lead in disks. (See below, pp. 898-902 . ) 



-331- 



L 
2 * 

4. 



Si! 



8 ; 

10! 

i 

i 

II ; 

i 

T -2 : ; 

14- ; 

I 

lS i 
1 

IS ; 

17 I 
IS : 
19 



■■\ 



20 'i 

1 

21 j 

-i 

^2. ;! 
1 

23 :| 

24. 

25 



manufacturers in this regard. 



Marketed By 


Mf g . By 


2314 Type Media 


Status 


Burroughs 


Burroughs 




No 


Fixed Disc 


Delivered 


CDC 


CDC 




Yes 


2316 


Announced FCS 2Q70 


DEC 


Memorex 




Yes 


2316 


Imminent Delivery 


GE 


IBM 




Yes 


2316 


Announced 


Honeywell 


Honeywell 




Yes 


2316 


Announced FCS 2Q70 


IBM 


IBM 




Yes 


2316 


Delivered FCS 1Q67 


NCR 


NCR 




No 


Strip 


Delivered 


RCA 


RCA 




Yes 


2316 


Announced FCS 1Q70 


SDS 


Memo rex/CDC 


Yes 


2316 


Imminent Delivery 


Univac 


Univac 




Yes 


2316 


Announced FCS 1Q70 



"The rental prices offered by CDC, GE, Honeywell, RCA, and Univac 
are within a few percentage points of the IBM 2314. (CDC and 
Honeywell discount by approximately 10% for three to five-year 
leases.) Burroughs and NCR use radically different approaches 
and price comparisons cannot be weighed properly due to the 
lower performance levels of their devices. To date, competitive 
system manufacturers have not had any significant price advan- 
tage in the file facility environment." (DX 7858, p. 2.) 

(viii) Standard Interface /Modularity. IBM adopted a 

"standard interface" for the peripherals in the compatible 360 line. 

This meant that (with some exceptions*) the same peripherals would 



* Such exceptions as existed came about as a result of design 
trade-offs. Some peripherals such as the 2301 and 2303 drums with 
high speeds, for example, were not made attachable to the slower 
models of System/3 60 (such as the Models 20, 22, 25 and 30) because 
those smaller CPUs could not accept the high data rates of these 
peripherals. (Case, Tr. 73449-50.) 

In some instances (such as with the 360/25) peripherals were 
attached directly to the CPU rather than through the standard inter- 
face because designing a "native attachment, closely integrated with 
the computer", provided "somewhat greater performance at somewhat 
lesser cost". (Hughes, Tr. 71941; Case, t Tr. 73450; see also PX 
2209A, pp. 15, 17.) In such cases, of course, the cost/performance 
improvements were achieved at the expense of some of the configuration 
flexibility that was afforded by the standard interface. (Hughes, 
Tr. 71941-42, 71995.) The dilemma of when to make such trade-offs 
was a difficult one both during the development stages of System/36 
(see Gardner, Tr. 38387-88, 38958-61, 39110-13; DX 1656, DX 1657, 
DX 1658, DX 1659) and thereafter. (See Haughton, Tr. 95019-24; 
DX 1662.) 



i r 



-332- 



1 
2 

3 

4 

5 

6 

7 

S 

* 
LQ 
11 
12! 
13 
14 
15 

16 . 

17 ! 
13 = 



attach to ail processors in the line and would do so in the same 
way. The standard interface, together with compatibility, helped 
maximize the benefits that customers could derive from the broad 
range of peripherals offered with 360 and the compatibility across 
the entire line. It helped give System/360 a configurability that 
was unmatched by competitors and permitted customers the utmost 
flexibility to optimize their data processing systems by piecemeal or 
modular changes. At the same time, it enabled IBM to reduce costs 
through economies in development and manufacturing. Others 
undoubtedly recognized these benefits and also moved toward more 
modular product lines — but not until well after IBM had done so. 
(Case, Tr. 73446, 73474-75, 73523.) 

The requirement for a standard interface for the New Product 
Line was implied by two of the architectural and engineering "ground 
rules" set out in the SPREAD Report — i.e., that "all channels shall 
appear identical" to any I/O device type and that -"the I/O gear shall 
not need to be changed" when one processor is substituted for a 
slower one. (DX 1404A, pp. 19-20 (App. A to JX 38).) Case and 



iq :| Hughes testified that the standard interface became a "design objec- 

2Q :j tive" for System/360. (Hughes, Tr. 34102-03; Case, Tr. 73446.) Case 

1 

3«j I explained: 

22 ■■ "[W]e had as an objective to design a number of different peri- 

•i pheral devices that would each be able to plug into central 
2- ! processing units of the whole System/36 family. We wanted to 

| do this in a way which would maximize the degree of choice that 
* A \ customers would have in selecting peripheral devices to go with 

""* j central processing unit models, and to do it in a way which 
m S j would minimize IBM's development expenses in designing those 



-333- 



L 
2 

3 

4-! 

5 

7 
3 

ia 

Li 
12 

14 



peripheral devices, and do it in a way which would help us to 
reduce our manufacturing costs of the peripheral devices by 
achieving as large as possible a production run of each par- 
ticular device. 

"The technique that we chose to accomplish these objectives 
was called the System/360 channel to control unit interface, 
often abbreviated with the words 'standard interface 1 ". (Case, 
Tr. 73446.)* 

The standard interface, together with compatibility, provided 

IBM with a number of development and manufacturing advantages. "It 

reduced the design time of many groups" who would otherwise have 

spent time designing their "own pet means of attachments". (Hughes, 

Tr. 71939.) Instead, the CPU and peripherals designers were able to 

concentrate on building "the best products they knew how" and on 

"advancing the state of their art as far as possible". (Case, Tr. 

73447.) 

The standard interface, together with compatibility, also 

helped IBM reduce ' development costs by reducing the number of circuits 

that had to be designed to permit each peripheral to attach to each 

j CPU. Prior to System/360, peripherals that attached to the central 

j processing unit did so by means of a unique interface. As a result, 

[ a separate design effort and set of circuitry was required for each 

such attachment to the central processing unit. With much of System/ 

360, only a single design effort and set of circuits was required 



17 

is i 

r 

ZQ 'i 

.1 
21 ■! • ! 

j * The control unit to peripheral device interface was not stan- •; 

23 ! dardized, however, which meant that each device required its own con- • 

j trol unit. The objective of the New Attachment Strategy in the 1970s 

2i, ; was to standardize the device to control unit level interface and j 

I thereby achieve benefits similar to what had been obtained with the ; 

?q I standardization of the control unit to channel interface in System/360. ! 

! (Case, Tr. 74079-83; Haughton, Tr . 95010-32.) " j 

! i 

i ' 

1 ' 

'■\ -334- ! 



because of the standardization of the interface between the control 

unit and the channel of the central processing unit. (Case, Tr. 

73446-48.) 

The standard interface, together with compatibility, helped 

simplify and cost reduce IBM's manufacturing process. "[I]t led to 

higher quantity production runs of the peripheral devices since the 

same peripheral device and the same attachment, or plug-in circuitry, 

was associated with the interface to any of the CPU models". (Hughes, 

Tr. 71939-40; Case, Tr. 73448.) Because of this commonality, similar 

economies were achieved in the testing process. That was particularly 

important to IBM in getting 360 ready for announcement. Hughes 

testified that 

"since we had a multitude of I/O devices and a prescribed time 
to get it done, [compatibility and the standard interface] helped 
us a great deal in both our engineering and all aspects of 
testing ... to get the total job done". (Tr. 71939-40; see 
also Case, Tr. 73533.) 

Case testified that a related objective of the 360 Advanced 

I Systems Group was to develop "elements of a computer system which 

| could be put together, or configured in a wide variety of ways". 

I 

| (Tr. 73416.) That objective, which Case called "modularity", was 

i 

:j promoted by the standard interface because it allowed users to plug 

1 any peripheral device into different 3 60 central processing units 

! 

:j "without changes in the central processing unit". (Case, Tr. 73448; 

i 

j see also Hughes, Tr. 34109.) 

i Not only did IBM achieve the modularity objective set for 

.j 

| System/36 (Case, Tr. 73420) , it did so to an extent that other manu- 

I facturers were unable to match for almost a decade. Among the manu- 
i 

I -335- 



L ; 
2 

3 

4- 

5 

a 

7 

3 

9 
10 
II 
IZ 
13 

14. 

« ^» 

15 



?T 



facturers and marketers of computer systems from 1964 to 1972, 

"IBM was the leader in providing . . . modularity. [*] With the 
announcement of the System/360, IBM provided the first line 
offering anything like the degree of modularity which has since 
become available from all the major manufacturers. 

"During the 1960's, all of the manufacturers, including IBM, 
evolved their product lines further in the direction of making 
them more modular, but ... it is fair to say that throughout 
the period . . . IBM's product line remained the most modular 
of all the general purpose product lines available". (Withington, 
Tr. 58268-69.) 

Accordingly, System/360 's modularity provided benefits to users that 

were unavailable from competitors and provided an incentive to acquire 

36 0s that did not exist with respect to competitive systems. As Case 

testified: 

"The achievement of the modularity objective was . . . very 
helpful to IBM in enabling the computer products produced by 
IBM to be chosen by customers in a way that would optimize 
the price/performance of their installation, and in a way which 
would provide for convenience and small accepted changes in 
the installation as the requirements of the enterprise changed. 

"That is an important benefit to customers for two reasons: 



"First, . . . they can most accurately adjust the capabili- 

] ties of their computing installation and, hence, the cost to ' . 

17 ; them of their computing installation to their real needs. 

XS I "Second, . . . they are able to change the performance or the 

;j capabilities of their configuration to match their changing 

£g i requirements . . . without changing the entire installation, but 

;j just adding or subtracting parts, or boxes from the installation." 

?0 "• (Tr. 73427-28; see also Navas, Tr. 41394-95; Withington, Tr. 

! 56193.) 



* " [A] modular line of computer systems is one in which every element; 
22 \ of the system, including processor, storage, peripheral equipment, and j 
systems programs can be independently exchanged for a compatible larger I 
or successor module in such a manner that over time the installed com- j 
puter system may evolve to a much different or a much larger or a much : 
more capable one without any particular point in time being identifiable! 
as one in which the entire system was converted from one to another." i 
(Withington, Tr. 58268, see also Tr. 58269-76.) I 






-336- 



IBM's achievement of modularity for System/360 "helped 
to remove limitations on the use of computing equipment that had pre- 
viously existed" because it relieved users of the need to make "system" 
. changes. (Case, Tr. 73435-37. ) IBM, more than any other firm,* 
reaped the benefits of user demand for modular acquisition alterna- 
tives : 

"Because the achievement of the modularity objective was useful 
for customers, it was of benefit to IBM in that.it tended to 
increase the value of IBM products as compared to the products 
of others, and with an increased value, our sales tended to 
increase and that was important in the achievement of the total 
success, or the total order rate for System/360 computers and 
the peripheral devices that were part of those computing systems." 
(Case, Tr. 73428.) 

There were, however, risks associated with modularity and 

the standard interface. The design trade-offs necessary to create a 

system which could be assembled in a wide range of configurations, 

might have resulted in a design that was not optimal for any particular 

configuration, at a cost higher than it need otherwise have been. 

\ j Development of the standard interface entailed a similar risk "that 



no one attachment or no one plug-in capability [would be] optimal for 



: ; * Other companies followed IBM's lead in making their product lines 
■| more modular, but were not as advanced or fast moving. For examnle, 

<, '! 
■i ' 

! a) Modularity "was beginning to appear" in Honeywell's line 
I ! by approximately 1966, but it "was still far less than 

j available in the IBM line" and did not "span the range 

1 \ of available modular options that IBM's line did" through 

" j the 196 0s; 

| b) By 1977, Univac ' s line was "probably still deficient" 
i j compared to IBM; and 

s "\ c) "Burroughs' modularity was restricted by the narrowness of 
! its product line . . . through most of the 1960s". 

| (Withington, Tr. 53271-75.) 

! 

! -337- 



L| the particular device involved". (Case, Tr. 73531-32.) Thus, the 

2 : question of separate control units versus native attachment of peri- 

i 
2 ; pherals became a matter of some controversy within IBM, involving 

4*! important dissenters (such as Haanstra) from the stand-alone control 

J !• unit method of attachment which was finally adopted for most of 360. 

a!! (See DX 1656; DX 1657; DX 1658; DX 1659.) 

7 ij There was risk to IBM of another type as well. 360' s 

standard interface and modularity of design, together with its wide- 
ranging compatibility, presented an attractive target for competitors. 
The new, modular environment in which 360 would be offered created 
the prospect that other manufacturers would produce "modules" that 
would be marketed in direct competition with comparable IBM products . 
The standard interface of System/360 offered others the same advan- 
tages it gave IBM* — and more. As Case testified, 

"It reduces their design costs as it did for IBM, and it allows 
them to achieve higher production runs as it did for IBM, and 
it allows users to conveniently plug in peripheral devices of 
their manufacture just as it allows the convenient plug-in 
of devices of IBM manufacture". (Tr. 73474-75; see also Navas , 
Tr. 41395-96.) 



-0 I * That was particularly true because IBM published a number of manu- 

| als which were readily available "at a nominal charge of a couple of 
*- '{ dollars" and which described the mechanical, electrical and logical 

•j characteristics of IBM's interfaces in a way that permitted manufac- 
22 .j turers of peripheral devices to design "workable and safe" attachments 

! of their devices to an IBM channel and which permitted CPU manuf ac- 
-^ j turers to attach IBM peripherals to their own CPUs in a like manner. 

'(Shoemaker, Tr. 30867; Case, Tr. 74125-50; Peterman, Tr. 99441-43; DX 
2- 17590, Perkins, pp. 21, 24; DX 7591, Hilyer, p. 15.) IBM's OEMI 

| (Original Equipment Manufacturers Information) Manual for System/360 
25 jwas first made available in 1965. (Case, Tr. 74145.) 



-338- 



r 
i 



of being able to copy IBM's designs and use IBM's software without ! 
having to invest in developing either. As a consequence they could 
be expected to have lower costs than IBM and to offer their products 
at lower prices than IBM initially charged. (Case, Tr. 73523; Cary, 
Tr. 101333-37, 101339, 101374, 101629-31; see also Wright, Tr. 
J 13236-38; Enfield, Tr. 20765-68; G. Brown, Tr. 51812; Powers, Tr. 

95376-89, 95412-13, 95475-82; PX 3312A, p. R14; PX 3594A, pp. 4, 26, 36, 
40; PX 3681A, p. R-l; PX 4880, p. 3.) 

The prospect that others would be able to "tap" IBM's 
support and offer compatible products in competition with IBM was 
foreseen by the SPREAD Committee and others within IBM prior to 360 's 
announcement. (Knaplund, Tr. 90497-98; DX 1404A, p. 40 (App. A to 
JX 38); see also PX 3908A.) That prospect became a reality in the 
late 1960s and in the 1970s — with numbers of competitors offering 
replacements for each and every box in IBM's systems. IBM could not 
keep to itself the advantages of compatibility, modularity and the 
standard interface.* On the other hand, IBM really had little 



:j * John Navas of Memorex explained the benefits for a manufacturer 
' :j of plug-compatible products in being able to attach a single disk 
,| drive model to a variety of 360 processors: 



T ! 



t 



"From the standpoint of a company such as Memorex, it would 
tend to reduce product cost to minimize the number of models of 
•I a given type of disk drive which we would be producing. That 
- '.| would result in a higher production volume for each type of 

unit, and would result in less development expense associated 
with developing the various models .... 

"If Memorex had had to produce unique models of its 630 for 
each of the various models of IBM System/360 ... it would have 
probably increased the development expense, caused an increase 
in manufacturing costs, and increased the difficulty and adminis- 
trative expense associated with lease base management". (Tr. 
41395-96.) 

-339- 



L 
Z 

z 

4- 
f 

a 



alternative but to provide such features if 360 was to succeed. It 
was a matter of responding to "a competitive necessity". Because of 
user demand, "the manufacturers attempting to compete were forced to 
maintain continuous developments of different modular types of equip- 
ment that could be configured together". (Withington, Tr. 56174.) 

However, the great modularity of System/360 meant that IBM 



7 j would have to price each and every box in the system carefully.* 

8 : According to Knaplund, IBM had to make those prices attractive on a 

9 j box-by-box basis because users made box-by-box performance comparisons 

i 

IQ ; between IBM and its competitors; because System/360 was susceptible to 
1^ | such a wide range of configurations that a single box price that was 
12.1 out of line could make the whole system unattractive; and because 

< . 

12. '■■ competition was anticipated from suppliers of plug-compatible peri- 
!£ !( pherals and CPU's who would attempt to replace IBM's products on a 
15 ! box-by-box basis. (Knaplund, Tr. 90496-98.) That last reason, in 
^g 1 particular, made competitive box prices for System/360 "critical". 



« T i (Id.) 



19 1 



21 f 



j * IBM has always priced its products on a box basis, with each unit 
7- ; of EDP hardware (such as a CPU, tape drive, disk drive or terminal) 
"" j offered at a consistent price regardless of the type or number of 
~s 1 boxes that a user combined to configure his system. (Knaplund, Tr. 
*" :j 90495-500; Akers , Tr. 96665, 96675-76; Cary, Tr. 101386-37.) 



-340- 



21 \ . ( 



r 



Appendix 
Examples of System/360 Uses 

The following are some of the diverse applications for 
which System/3 60s have been used: 

By a French research and consulting firm to study ways of 
increasing the power output of large hydroelectric dams (DX 13677, 
p. 16) ; 

By a petroleum exploration company to prepare seismic 
reports ( id. , p. 14) ; 

By a manufacturer of animal feed concentrates for feed 
formulization (DX 13678, p. 9); 

By the Deutsches Elektronen Synchrontron in Hamburg, 
j Germany, to evaluate photographs of bubble traces left by invisible 
elementary particles in an electron accelerator (DX 13679, p. 20); 

By a Japanese steelmaker for automatic control of the 
steel manufacturing process ( id. ) ; 

By Swissair for automated message switching and automatic 



| passenger check-in and weight-and-balance calculating ( id. ) ; 

* 

\ By a paint manufacturer to signal corrections for 

* i 

:j deviations in ingredients and production cycle ( id. , p. 10) ; 

.1 By scientists in New England to simulate and study the life 

I '• 

i cycle of lobsters ( id. ) ; 

m 

j By African Ivory Coast harbor authorities to compile and 

! analyze statistics on tropical wood exports ( id. , p. 13) ; 

A \ 

j By a Swiss chemical manufacturer to operate an automated 



-Al- 



3 

s 

7 
3 

Q 

10 

u ; 

iz I 

! 
I 

n J 

L4* 
15 
15 



warehouse (DX 13680, p. 28); 

By BOAC to calculate tariffs, management statistics and 
flight plans ( id. ) ; 
4^i[ By Japan's national broadcasting company to maintain 

schedules and budgets for 640 television shows and 1,200 radio 
programs, and to control actual broadcasts ( id. , p. 16) ; 

By IBM's Field Engineering Division for computer assisted 
instruction (DX 3364, p. 9); 

By an air freight company for instantaneous tracking of 
daily shipments ( id. , p. 22) ; 

By the architectural department of a county council in 
England to design municipal buildings ( id. , p. 24) ; 

By a supermarket chain to calculate unit prices (PX 5767, 

il P- 13); 

By American Airlines (3 60/65) for airline reservations 
(Welch, Tr. 75385-86) , field maintenance reliability applications 

•| (O'Neill, Tr. 75848-53) , crew qualification and takeoff power assist 

17 

(id., Tr. 75909-10) , flight planning (id. , Tr. 75928) , and calcula- 



; tion of estimated time of arrival ( id. , Tr. 75976) ; 

i3 i 

•; By Aspen Computype, Inc. (3 60/4 0) for typesetting 

2C ■! 

'! (DX 6078, McCaffery, p. 9) ; 

2i "i . 

j By Autocomp, Inc., (360/40 and 360/50) for typesetting 

\ (DX 4039, Kendall, pp. 7-3); 

23 i , 

! By AVCO Computer Services m Wilmington, Massachusetts, 

~A \ 

*"" I (360/75) for: 



-A2- 



1 ! i 

2 !| 

3 j! 

5 ! 

1 

6 :f 

7 I 

ii 

8 j 

9 l 
10 j 

ii i 

12 | 

13 1 

.! 
.1 

14 || 

15 ii 

16 =i 

17 !: 

13 : ; 

19 :| 

20 '! 

21 ■ 

22 • 

23 i; 
24. ; 
25 ; 



drafting applications 

FORTRAN flowcharting 

geometric design 

mathematical functions 

frequency distributions 

movie making 

perspective plotting 

trajectory analysis 

financial analysis 
production control 

statistical analysis 

mathematical analysis 
applied statistics 

structural load analysis 
structural shell analysis 



structural ring and 
frame analysis 

antenna pattern 
prediction 

communication link 
analysis 

plasma attenuation 
analysis 

drag coefficient 
analysis 

aerodynamic heating 
analysis 

heat transfer 
analysis 

thermochemical 
equilibrium analysis 

flow field analysis 

boundary layer 
analysis 

penetration aids 
analysis 

decoy model analysis 

radar cross section 
analysis 

finance applications 

manufacturing 
applications 



(DX 6816, pp. 3, 10, 12, 13) ; 

By Bowne Timesharing, Inc. , (36 0/40 and 3 60/50) for time- 
sharing text editing (DX 6090, Abrams , pp. 9-10); 

By Carnation Corporation (360/40) for telecommunication 



-A3- 



L 
2 
3 

4- 

3 



5' 



7 
S 

ia 

Ll 
12 

13 

14 



is: 



17 i: 
is 



applications and linear programming (Navas, Tr. 39177-78); 

By Computone Systems, Inc., (360/50) for architectural 
design and mathematical modeling (DX 4069, Robeson, pp. 16-17); 

By Continental Illinois National Bank & Trust (360/50) for 
on-line credit authorization (DX 4756, p. 7); 

By DP&W, Inc., (360/30) for business and engineering 
applications (DX 4076, DiPietro, pp. 8-9); 

By the San Francisco Federal Reserve Bank (360/50) for 
message switching (Withington, Tr. 57540; DX 2667, p. 3); 

By the Fluor Corporation (360/50) for project planning and 
control, process simulation, process analysis, refinery simulation, 
structural design, piping design, electrical design and mechanical 
design (DX 4023, Neher, pp. 11-12, 17); 

By General Motors Research (360/50) for timesharing (Hart, 
Tr. 80505-08); 

By the New York Police Department (360/40) for automated 
dispatch and identification of police vehicles (DX 4756A, p. 58); 

By the Orange Coast College District in Costa Mesa, 
fcalifornia (360/50) for computer assisted instruction, grade report- 



1S ; 

ling and student registration (King, Tr. 14 761-62) ; 

;| By Pacific Southwest Airlines (360/65) for passenger service 

""applications (O'Neill, Tr. 76019); 

22 ''• 

i 3y Proprietary Computer Systems i Inc., (360/65) for: 

:j 

25 l - 

! 
I 
I 

1 

■! -A4- 



banking services 
accounting 

manufacturing control 
three dimensional COGO 
stress analysis 
digital signal processing 
reliability calculations 
electrical engineering 
fast fourier transforms 
matrix analysis 
chemical engineering 
graph plotting 



thermal analysis 

linear programming 

CPM analysis 

PERT analysis 

Monte Carlo analysis 

Markov analysis 

integration 

differentiation 

non-linear equations 

regression analysis 

descriptive statistics 



transducer calibration 
(DX 3960, Barancik, pp. 11-12); 

By Pyramid Industries, Inc., (360/40) for time shc.ring 
(DX 4756D, p. 23) ; 

By Southern Railway (360/50 and 360/65) for on-line 
monitoring of railroad cars (DX 4756D, p. 42) , (360/50) for peripheral 
processing (J. Jones, Tr. 79848, 79413-14), (360/30) for card to tape, 
tape to card, and tape to print processing, and peripheral processing 
( id. , Tr. 79243); 

By TBS Computer Centers Corporation (3 60/30 and 360/40) 
for data communication, remote teleprocessing, accounting reports and 
statistics, inventory, cost analysis, market research, production 
control, accounts receivable and payable, traffic studies and order 



-A5- 



L 
2 
2 

4- 
a 
a 
7 
3 
9 ! 

ia 

U ! 
12. 1 



analysis (DX 7134) ; 

By Union Carbide (3 60/30) for message switching (McGrew, 
Tr. 77271) . 

System/360 's uses within the Federal government alone 
illustrate graphically the broad range of applications performed by 
360 users. For example, 360s have been used: 

By the Headquarters, U*S. Marine Corps (360/20) , for 
"Automated Communications Processing System" (DX 2992,* pp. 619, 
1123-1125) ; 

By the Veterans Administration, Austin, Texas (3 60/20) , 
for "Patient Care" (DX 2992, pp. 1073, 1158); 

By the Veterans Administration, Washington, D.C. (3 60/20) , 
for "Facility Planning and Construction" and "Fiscal Accounting" 
: (DX 2992, pp. 1078, 1158) ; 

By the Veterans Administration, Philadelphia (360/20) , for 

"Insurance" (DX 2992, pp. 1076, 1158); 
IS ; i 

By the Defense Nuclear Agency, Headquarters, Field Command 



14- 



17 
15 
19 



(360/20) , for "Logistics - Supply" and "Stockpile Management" (DX 
2992, pp. 546, 1121; DX 4593, p. 133) ; 



\ By the Department of Air Force, Air Force Systems Command, 

axil 

Los Angeles, California (360/20), for "Telecommunications" and 



i 

fT .1 ' 

— -• ! 



: ] "Command and Control" (DX 2992, pp. 452, 1120; DX 4593, pp. 103, 104) 

'i 



23 \ 

J * DX 2992 is the Stipulation and Amended Response of Plaintiff to 
24. 'J IBM's Interrogatory 5(e). Examples of applications taken from DX 

j 2992 are described here in the same terms in which they are described 
25 j in DX 2992. 

■i - -A6- 



f i 
i 

» ' 
i ! 






By the Atomic Energy Commission, Division of Technical 
Information (360/20) for "Operations Control and Support" (DX 2992, 
pp. 118, 1113; DX 4593, p. 72); 

By the Atomic Energy Commission, Oak Ridge Office (360/20) , 
for "Scientific and Engineering" (DX 2992, pp. 91, 1113; DX 4593, 
| p. 61); 

By the- Department of Commerce, Office of the Secretary 
(360/20), for "General Administration" (DX 2992, pp. 149, 1117; 
DX 4593, p. 74) ; 

By the Department of Commerce, Bureau of the Census (360/20) , 
for "Statistical Programs" (DX 2992, pp. 157, 1117; DX 4593, p. 77); 

By the Department of Air Force, Air Force Systems Command, 
Eglin AFB (360/20), for "Research, Engineering" (DX 2992, pp. 442, 
1120; DX 4593, p. 101) ; 

By the Department of Air Force, Air Force Communications 
Service, Offutt AFB (360/20), for "Weather, Environment"" (DX 2992, 
pp. 411, 1120; DX 4593, p. 94); 

By the Marine Corps Headquarters, FMFLANT (360/20), for 
"Automated Communications Processing System" (DX 2992, pp. 6 31, 
1123-25) ; 

By the Department of Navy, Naval Intelligence Command 

(360/20), for "Intelligence Data Handling System" (DX 2992, pp. 733, ! 

I 
1123-25; DX 4593, p. 156); | 

I 

By the Department of Navy, Naval Research Laboratory j 
(360/20), for "Laboratory Support Systems" (DX 2992, pp. 610, 1123-25) ;j 



-A7- 



L 

2! 
3 
±\ 

a 
7 
3 

ia 
u 

12 
12 

14 
15 

IS 
17 
IS 



By the Department of Navy, Commander Naval Reserve 
(360/20) , for "Navy Manpower and Personnel Management Information 
System" (DX 2992, pp. 657, 1123-25; DX 4593, p. 138); 

By the Department of Navy, Pacific Fleet Commander in 
Chief (360/20), for "Air Logistics Support Systems" (DX 2992, pp. 
722, 1123-25; DX 4593, p. 152); 

By the Defense Supply Agency, Lemoncove, California 
(360/20), for "Communications" (DX 2992, pp. 802, 1126); 

By the Export/Import Bank of the U.S. (360/20), for 
"Payroll and Personnel", "Accounting" and "General Administration" 
(DX 2992, pp. 818, 1127); 

By the National Aeronautics and Space Administration, 
Goddard Space Flight Center, Greenbelt, Md. (360/20) , for 
: "Scientific" and "Engineering" (DX 2992, pp. 907, 908, 1144); 

By the National Aeronautics and Space Administration, Jet 

Propulsion Laboratory, Pasadena, California (360/20) , for "Business- 
i 

j Commercial" (DX 2992, pp. 937, 1144); 

| By the Department of Treasury, Office of Treasurer (360/20) 

I for "Administration of Government Finances" (DX 2992, pp. 1066, 1155; 



| DX 4593, p. 194) ; 

ZQ "I 

! By the U.S. Defense Communication Agency (3 60/20 and 

21 '• 

j 360/30), for communication control and as terminals (DX 7524, 

22 '[ 

j Levine, pp. 34-36, 57); 

23 '; 

I By thts Atomic Energy Commission, Brookhaven National 

i Laboratory (360/30) , for "Personnel Management" and "Operations 



-A8- 



Control and Support" (DX 2992, pp. 6, 1113); 

3y the Civil Aeronautics Board (36 0/30) for "Traffic 

Capacity", "World Benefit Study", "Air Cargo" and "Payroll, Manpower 

Distribution" (DX 2992, pp. 134, 1116); 

» 

By the Department of Army, Air Defense Board (360/30) , 
for "Research, Engineering" (DX 2992, pp. 174, 1120); 

By the Department of Air Force, Air Force Finance Center 
(360/30), for "Finance, Accounting", and "Payroll, Benefits" (DX 2992, 
pp. 279, 1120) ; 

By the Department of Air Force, Air Force Systems Command 
(360/30), for "Research, Engineering" (DX 2992, pp. 433, 1120; 
DX 4593, p. 99) ; 

By the Defense Communications Agency, European Area 
(360/30), for "Communications Control and Management" (DX 2992, pp. 
550, 1122; DX 4593, p. 133); 

By the Defense Nuclear Agency, Headquarters Field Command 
(360/30), for "Test Command", "Accounting and Finance", "Communica- 
tions Processing" and "Data Automation" (DX 2992, pp. 54 6-47, 1121; 

DX 4593, p. 133) ; 

: ! 

' ;; By the Department of Labor, Bureau of Labor Statistics 

} ,j (360/30), for "Statistical and Economic Survey Appl." (DX 2992, 

:i 
L ! pp. 884, 1142; DX 4593, p. 175); 

:■! 
- :j By the National Aeronautics and Space Administration, 

5 Johnson Space Center (360/30), for "Scientific" and "Business- 



Commercial" (DX 2992, pp. 971, 1144); 



-A9- 



L j 
Z.\ 
z ! 

3 if- 

a 

7 

3 

9 
10 
li 
12 
13 
!A 
15 
la 
17 

i 
IS 



20 ! 

•i 

21 I 

•I 
22 

23 i 
2± j 

! 



By the National- Aeronautics and Space Administration, 
Goddard Space Flight Center (360/30) , for "Mission Control" and "Data 
Reduction" (OX 2992, pp. 907-8, 1144); 

By the Tennessee Valley Authority, Computing Center Branch 
(360/30), for "Power Supply and Use", "Fertilizer and Munitions 
Development", "Financial Management" and "Personnel Management" (DX 
2992, pp. 1068, 1156; DX 4593, p. 194); 

By the Securities and Exchange Commission, Office of Data 
Processing (360/30), for "Mass Information Storage and Retrieval", 
"Statistical and Economic Analyses" and "Administrative Processing 
(Personnel, Payroll, etc.)" (DX 2992, pp. 1025, 1151, DX 4593, p. 191); 

By the Department of Transportation, Transportation Systems 
Center (360/30), for "Financial Administration", "Inventory, Supply 
and Logistics" and "Planning, R and D" (DX 2992; pp. 1034, 1154; 
DX 4593, p. 165) ; 

By the Social Security Administration (360/30) for real 
time claims tracking and real time tape library control (DX 5 792, 
pp. 17-18); 

By the Department of Navy, Marine Corps - COMCAB West 
(360/30), for "Automated Communications Processing System" (DX 2992, 
pp. 571, 1123-25); 

By the Department of Navy, Chief of Naval Operation (36 0/30) , 
for "Space Surveillance System" (DX 2993, pp. 669, 1123-25; DX 4593, 
p. 141); 

3y the Department of Navy, Ordnance Systems Command (36 0/30) , 



-A10- 



for "Ordnance Support System" (DX 29 92, pp. 694, 1123-25; DX 4 593, 
p. 145) ; 

By the Defense Supply Agency, Assistant Director of Plans 
(360/30), for "Logistics - Depot Level" and "Communications" (DX 2992, 
pp. 812, 1126; DX 4593, p. 135); 

By the Defense Supply Agency, Assistant Director of Plans 
(360/30), for "Headquarters Management" (DX 2992, pp. 815, 1126; 
DX 4593, p. 136) ; 

By the Federal Deposit Insurance Corporation, Division of 
Research (360/30), for "Economic Research" ,. "Fiscal Accounting", 
"Bank Liquidation" and "Bank Statistics" (DX 2992, pp. 823, 1130; 
DX 4593, p. 171) ; 

By the Department of Health, Education, & Welfare, Food and 
Drug Administration (360/30) , for "Disease Prevention and Control" and 
"Consumer Protection" (DX 2992, pp. 855, 1137; DX 4593, p. 173); 

By the Atomic Energy Commission, Albuquerque Office 

(360/40), for "Material Management", "Facilities Management", 

"Operations Control and Support" and "Scientific and Engineering" 

(DX 2992, pp. 35, 1113; DX 4593, p. 46); 
i 

I By the Department of Commerce, National Oceanic and 

n '; 

* I Atmospheric Administration (360/40), for "Mapping, Charting and 



a 
7 



j Marine Description" (DX 2992, pp. 148, 1117; DX 4593, p. 74); j 



^ i 



By the Department of Commerce, National Oceanic and 

i- i 

a j Atmosphere Administration (360/40), for "Environmental Prediction 
j 

•^ j and Warning" (DX 2992, pp. 150, 1117; DX 4593, p. 74); 



-All- 



L 
Z 
3 

f 

7 
3 
9 

io ; 
ii 

12 

13 



By the Department of Commerce, Office of Administration 
Domestic International Business (360/40), for "Statistical Programs", 
"Economic Analysis" and "Industrial Mobilization" (DX 2992, pp. 157, 
1117; DX 4592, p. 46) ; 

By the Department of Army, U*.S. Army Munitions Command 
(360/40), for "Research, Engineering" (DX 2992, pp. 169, 1120; DX 4593, 
p. 126); 

By the Department of Air Force, Air Force Logistics Command 
(360/40), for "Supply, Inventory Control, Cataloging" (DX 2992, pp. 273, 
1120; DX 4593, p. 87) ; 

By the Department of Air Force, Aerospace Defense Command 
(360/40), for "Telecommunications", "Command and Control", 
"Intelligence" and "Tracking" (DX 2992, pp.418, 1120; DX 4593, p. 96); 

. . t By the Department of Air Force, Air Force Systems Command 

I* ; 

i 

; (360/40) , for "Research, Engineering" (DX 2992, pp. 428, 1120; DX 4593, 

ls . 

P- 98); 

la 

j By the Department of Transportation, FFD Aviation Administra-- 
17 :j 

jtion (360/40), for "Inventory, Supply and Logistics", "Mission 

- 3 ; 

■Support, Operations" and "Planning, R and D" (DX 2992, pp. 1039, 1154; 

19 : ! 

JDX 4593, p. 166) ; 

20 ] 

i By the Office of Economic Opportunity, Office of the 

21 -I 

.{Comptroller (360/40), for "Payroll Accounting", "Personnel Accounting" 

22. '- . 

jand "Research and Development" (DX 2992, pp. 1015, 1146; DX 4593, 

23 



|p. 191); 



i* 



15 



By the Securities and Exchange Commission, Office of Data 



-A12- 



1 ij Processing (360/40), for "Mass Information Storage and Retrieval", 

2 ;! "Statistical and Economic Analyses" and "Administrative Processing 



3 

9 

10 

11 
12 



(Personnel, Payroll, etc.)" (DX 2992, pp. 1025, 1151? DX 4593, p. 191) 
By the Veterans Administration, Department of Data Manage— 
- ij ment (360/40), for "Loan Guaranty" and "Facility Planning and 
6 !| Construction" (DX 2992, pp. 1073, 1158; DX 4593, p. 195); 

t 

_ :| By the Veterans Administration, Department of Data Manage- 

ment (360/40), for "Patient Care" (DX 2992, pp. 1078, 1158; DX 4593, 
p. 196); 

By the Department of Navy, Director of Naval Laboratories 
(360/40), for "Laboratory Support Systems" (DX 2992, pp. 685, 1123-25; 
DX 4593, p. 143) ; 

<\ By the Department of Navy, Air Systems Command (360/40) , 

13 j 

';! for "Air Logistics Support Systems" (DX 2992, pp. 565, 1123-25; 

14 !j 

ij DX 4593, d. 139) ; 

15 ij 

!i By the Federal Deposit Insurance Corporation, Division of 

15 : ' 

:i Research (360/40), for "Economic Research", "Fiscal Accounting" and 
17 

18 

19 

20 

21 

22 

23 

24 

25 



"Bank Statistics" (DX 2992, pp. 823, 1130; DX 4593, p. 171); 

By the Government Printing Office, Assistant Public Printer j 
(360/40), for "Inventory Accounting and Control" and "Electronic 

Printing" (DX 2992, pp. 834, 1135; DX 4593, p. 171); 

I 
i 

By the Department of Health, Education and Welfare, Food ! 

i 

and Drug Administration (360/40), for "Disease Prevention and Control"! 

i 

(DX 2992, pp. 855, 1137; DX 4593, p. 173); j 

i 

l 

By the Defense Nuclear Agency, Headquarters, Field Command 



-A13- 



1 
2 
3 

4* 

3 

7 
8 
9 

10 

LI 

rz 

13 
14 
IS 
15 ,. 
17 i| 

ra I 

~ : ! 
* \ 

\ 

•i 



(360/40), for "Accounting and Finance", "Nuclear Weapons Materiel Con- 
trol", "Stockpile Management" and "Test Command" (DX 2992, pp. 546, 
1121; DX 4593, p. 133) ; 

By the Department of Navy, Marine Corps (360/40) , for 
"Manpower Management System" (DX 2992, pp. 615, 1123-25; DX 4593, 
p. 159); 

By the Department of Navy, Pacific Commander-in-Chief 
(360/40), for "Intelligence Data Handling System" (DX 2992, pp. 753, 
1123-25; DX 4593, p. 163); 

By the Air Force Aeromed Installation (3 60/40) to simulate 
bombing equations, radar signal acquisition and airborne computers 
(DX 5640, Mayer, p. 34); 

By the Army (van-mounted 360/40s) for maintaining a running 
account in the field of supply and demand of field support services 
(Wright, Tr. 13394-95; DX 913); 

By the U.S. Army Strategic Communications Command (360/40) 
for message switching (Wright, Tr. 13412-13) ; 

By the Atomic Energy Commission, Chicago Office (360/44), 
for "Material Management", "Financial Management" and "Scientific and 
Engineering" (DX 2992, pp. 84, 1113; DX 4593, p. 60); 

By the Department of Air Force, Air Force Systems Command 
(360/44), for "Research, Engineering" (DX 2992, pp. 284, 1120); 



££ 



By the Department of Air Force, Strategic Air Command 
I (360/44) for "Command and Control" (DX 2992, pp. 319, 1120? DX 4593, 
.! p. 80)'; 



-A14- 



L 
I 

I 

4- 

J 

/ 
3 

a 



3 ' 



By the National Aeronautics and Space Administration, Office 
of Manned Space Flight (360/44), for "Simulation" (DX 2992, pp. 984, 
1144; DX 4593, p. 188) ; 

By the National Aeronautics and Space Administration, 
Advanced Research' and Technical Office (360/44) , for "Test Data. 
Acquisition" (DX 2992, pp. 904, 1144; DX 4593, p. 177); 

By the National Aeronautics and Space Administration, Flight 
Research Center, Edwards Air Force Base, California (360/50) , for 
"Scientific", "Engineering", "Data Reduction" and "Business Commercial" 
(DX 2992, pp. 905, 1144) ; 

By the Railroad Retirement Board, Data Processing and 
Accounts Bureau (360/50), for "Research and Actuarial Services", 
"Process of Unemployment and Sickness Benefits" (DX 2992, pp. 1021, 
1149; DX 4593, p. 191) ; 

By the Tennessee Valley Authority, Computing Center Branch 
(360/50), for "Resource Development and Management", "Power Supply and 
Use", "Fertilizer and Munitions Development" and "Personnel Management". 
(DX 2992, pp. 1068, 1156; DX 4593, p. 194); 

By the Department of Transportation, Federal Highway Adminis- 
tration (360/50), for "Inventory, Supply and Logistics", "Planning, 
R and D" and "Mission Support, Operations" (DX 2992, pp. 1050, 1154; 
DX 4593, p. 170) ; 

By NASA's Flight Research Center in Edwards, California 

23 '! 

J (360/50), for reduction and analysis of flight data, scientific 

?A ■! 

* J theoretical calculations and administrative data processing (DX 5 308, 

:« I 

- ; !p. 1); 

:| -Ai5- 

■i 



.6 
.7 ; 

,3 : 

it 
i 

.i 

7", .1 



~ J 

: i 



L 
2 

3 

4j 



By NASA's Kennedy Space Center (360/50) for real time 
inventory management, integrated launch, vehicle modification status, 
payroll and remote file inquiry (DX 5256, pp. 6, 63); 

By the U.S» Coast Guard and Geodetic Survey. Office (360/50) 
for developing aeronautical charts , analyzing satellite data , provid- 
ing tidal data, locating earthquakes and assisting in geomagnetic 
7 ';{ studies (Wright, Tr. 13410-12; DX 13678, p. 9); 

By duPont's Savannah River Laboratory Plant (360/50) for 
neutron the realization and reactor kinetics (H. Brown, Tr. 83244-49); 

By the Department of Navy, Commander-in-Chief Pacific Fleet 
(360/50), for "Intelligence Data Handling System" and "CINCPAC Support 
Information System" (DX 2992, pp. 752, 1123-25; DX 4593, p. 162); 

By the Department of Navy, Commander-in-Chief Pacific Fleet 
(360/50), for "Material Management Information System" (DX 2992, 
pp. 714, 1123-25; DX 4593, p. 148); 

By the Department of Navy, Facilities Engineering Command 
(360/50), for "Ordnance Support Systems" (DX 2992, pp. 717, 1123-25; 
DX 4593, p. 148) ; 

By the Department of Navy, Air Systems Command (360/50) , 
for "Air Logistics Support Systems" (DX 2992, pp. 658, 1123-25; 
DX 4593, p. 138) ; 

By the Federal Deposit Insurance Corporation, 
Division of Research (360/50), for "Economic Research", 



I ! 



By the Government Printing Office, Assistant Public Printer 
(360/50), for "Payroll, Earnings and Leave Accounting", "Electronic 
Printing" and "Inventory Accounting and Control" (DX 2992, pp. 834, 
L I 1135; DX 4593, p. 171); 

By the Department of Health, Education and Welfare, Food 
^ | ; and Drug Administration (360/50), for "Disease Prevention and Control" 
and "Consumer Protection" (DX 2992, pp. 855, 1137; DX 4593, p. 173); 

By the Atomic Energy Commission, Albuquerque Office (360/50) , 
for "Facilities Management", "Operations Control and Support" and 
"Scientific and Engineering" (DX 2992, pp. 35, 1113; DX 4593, p. 46); 

By the Department of Army, White Sands Missile Range, New 
Mexico (360/50), for "Research, Engineering" (DX 2992, pp. 177, 1120); 

By the Defense Communication Agency, NMCS Support Center 
(360/50), for "Gaming, Modeling, and Systems Development", "Command 
and Control" and "Damage Assessment" (DX 2992, pp. 551, 1122; DX 
4593, p. 133) ; 

By the Department of Air Force, Sacramento Air Material 

Area, McClellan Air Force Base, California (360/50), for "Personnel" 

8 jj 
,and "Education" (DX 2992, pp. 273, 1120); 

9 . . i 
| By the Department of Air Force, Aeronautical Systems Division, j 

fl i ■ ! 

Wright Patterson Air Force Base, Ohio (360/50) , for "Research, Engineer- \ 

1 • 

!ing" (DX 2992, pp. 282, 1120); 

~ 1 I 

:j By the Department of Air Force, Pacific Air Force (36 0/50) , ! 

2 | I 
•I for "Command and Control" (DX 2992, pp. 456, 1120; DX 4593, p. 105); j 

i By the Department of Air Force, Strategic Air Command j 

"1(360/50), for "Intelligence" (DX 2992, pp. 322, 1120; DX 4593, p. 80); j 



J 

r 

■ 

y i 
i 

1 

m 

6 ( 

7 I 



i 

ii -A17- 



1 
2 

2. 

4* 

6 

7 

3 

9- 
10 
II 
12 
13 
14 
15 

is ; 

17 

is 

19 ; l 



By the Atomic Energy Commission, Savannah River Office 
(360/65), for "Material Management", "Financial Management" and 
"Scientific and Engineering" (DX 2992, pp. 115, 1113; DX 4593, p. 71); 

By the Department of Army, Safeguard, Whippany, N.J. 
(360/65), for "Research, Engineering" (DX 2992, pp. 165, 1120); 

By the Department of Air Force Ogden Air Material Area, 
Ogden, Utah (360/65), for "Payroll, Benefits", "Procurement, Contract 
Administration" and "Law Enforcement" (DX 2992, pp. 275, 1120); 

By the Department of Air Force, Air Force Systems Command 
(360/65), for "Research, Engineering" (DX 2992, pp. 438, 1120; DX 
4592, p. 67); 

By the Department of Air Force, Aerospace Defense Command 
j (360/65), for "Command and Control" (DX 2992, pp. 418, 1120; DX 4593, 



p. 96); 

By the Department of Air Force, Air Force Systems Command 
(360/65), for "Intelligence" (DX 2992, pp. 428, 1120; DX 4593, p. .97).; 

By the Defense Communications Agency, NMCS Support Center 
(360/65), for "Gaming, Modeling, and Systems Development", "Command 
and Control" and "Damage Assessment" (DX 2992, pp. 551, 1122; DX 4593, 



;|P- 133); 
20 i 



By the Department of Navy, Marine Corps Automated Service 



21 •'! 

jCenter, Kansas City, Missouri (360/65), for "Manpower Management 

ISystem" and "Personnel Accounting System" • (DX 2992, pp. 628, 1123-25); 

23 I 

' By the Department of Navy, Naval Air Development Center, 

** -iwarminster, Pennsylvania (360/65), for "Laboratory Support Systems" 
25 j 



(DX 2992, pp. 568, 1123-25); I 

i 

-A18- 



L i By the Department of Interior, Geological Survey (360/65) , 

2 l for "Recreation Use and Preservation" (DX 2992, pp. 877, 1140; 
DX 4593, p. 174) ; 

By the Department of Labor, Departmental Data Processing 
Center (360/65) , for "Accounting and Payroll Services" and for 
"Statistical Data Gathering" (DX 2992, pp. 883, 1142; DX 4593, p. 175) 

By the National Aeronautics and Space Administration, 
Goddard Space Flight Center (360/65) , for "Scientific" and "Engineer- 
ing" (DX 2992, pp. 908, 1144); 

By the National Aeronautics and Space Administration, 
Goddard Space Flight Center (360/65), for "Data Reduction" (DX 2992, 
pp. 907, 1144); 

By the National Aeronautics and Space Administration, 
>, j| Johnson Space Center (360/65), for "Simulation" (DX 2992, pp. 983, 
, s \ 984, 1144) ; 

By the Tennessee Valley Authority, Computing Center Branch 
(360/65), for "Power Supply and Use", for "Fertilizer and Munitions 
Development" and for "Employee Health and Safety" (DX 2992, pp. 106 8, 
1156; DX 4593, p. 194) ; 



4- 

a 
7 
3 
9 

a 



L6 ! 
IS 



:a1 

j By NASA's Johnson Space Center (360/65) for Sky lab 

i simulation (DX 7536, Woodling, pp. 23-24) ; 

21 '*. 

j By the Navy Computer Sciences Department in San Diego 

22 ;i 

I (360/65) for processing complex scientific and management type data 

| and for time sharing (DX 5100, pp. 17, 28); 

Z- ! 

| By the Air Force Eastern Test Range (3 60/6 5) for mechanized 



-A19- 



range scheduling, radar data reduction, trajectory measurement, optical 
infrared system data reduction (DX 5023, pp. 1-4); 

By the California Institute of Technology's Jet Propulsion 
• Laboratory (360/75) for real time mission control, simula- 
tion and real time telemetry (DX 5296, pp. 4, 6, 7); 

By the Naval Electronics Laboratory Center in San Diego 
(360/65) for interactive time sharing (DX 4334, pp. 1, 5) ; 

By the Department of Air Force, Air Force Systems Command 
(360/67), for "Telecommunications" and "Command and Control" (DX 2992, 
pp. 451, 1120, DX 4593, p. 103); 

By the Defense Communications Agency, NMCS Support Center 
(360/67), for "Command and Control" and "Damage Assessment" (DX 2992, 



14- ;' 



By the Department of Navy, Post Graduate School (3 60/67) , 



. ! for "Management Information System for Education and Training" (DX 

U3 



2992, pp. 588, 1123-25) ; 

By the National Aeronautics and Space Administration, 
Ames Research Center (360/67), for "Scientific" and "Business- 



IS 

17 

i Commercial" (DX 2992, pp. 888, 1144); 

I By the Atomic Energy Commission, Idaho Office (360/7 5) , for 

20 ] 

'l "Material Management", "Financial Management", "Personnel Management" 

21 4 . 

"j and "Operations" (DX 2992, pp. 77, 1113; DX 4593, p. 58); 

22 ;! 

;j By the National Aeronautics and Space Administration, 

23 ; 

I Goddard Space Flight Center (360/75), for "Scientific", "Engineering" 

2* 1 

j and "Mission Control" (DX 2992, pp. 907, 908, 1144); 

25 ! 

j 

i -A20- 

•i 

■i 

r 

•I 

'i 
it 



LI 



By the National Aeronautics and Space Administration, Jet 
Propulsion Laboratory (360/75), for "Data Reduction" (DX 2992, pp. 
I I 946, 1144); 



i- 
i . 

f r 

z. 
7 
S 

a 

i 
i 

A 

«* 

.5 .1 

■7\ 

3 ; 

"* .1 



I 
LI =! 

\ 
■2. ^ 

i 
.- ii 



'i 



-A21- 



1 35. The System/360 Commitment . System/360 was a "fantastip 

2 • undertaking" involving "fantastic risks". (cary, Tr. 101359; see 

3 j also Brooks, Tr. 22868; Case, Tr. 73561; Evans, Tr. 101126.) 360 was 
^ : "vastly different" from anything IBM had previously undertaken in 

5 : terms of "magnitude, complexity and functional characteristics", and 

6 • was "fundamentally new and different" compared to competitors' EDP 

7 offerings as well. (Knaplund, Tr. 90515; Evans, Tr. 101126; PX 1092 

8 p. 1; DX 117'2, pp. 1-2.) It was clear from the outset that no half- 

9 • way measures would suffice to carry out the SPREAD Committee's plans— f 

10 | and non was taken. IBM committed more "skill and energy" and 

11 "corporate resources" to the successful implementation of System/360 

12 than to any previous undertaking in its history. (PX 1900, p. 4.) 
Virtually the whole IBM's EDP operations were involved in 

the development and manufacture of System/360. The scope and magni- 
tude of the undertaking required a worldwide, interdivisional effort 
on IBM's part. "From its inception, System 360 was designed, dev- 

17 i eloped and tested for worldwide use, and was in fact used worldwide". 

18 I (McCarter, Tr. 88377; DX 1404A, p. 8 (App. A to JX 38).)* 
19 

20 *The 360/30 was developed in Endicott and was manufactured in 
Endicott, Sindelf ingen, Germany, and Mainz, Germany. (Dunlop, Tr. 

21 93647.) The 360/40 was developed in Hursley, England, and manufac- 
tured in Poughkeepsie, Essones, France, and Montpellier, France. 

22 (Id.; Hughes, Tr. 33921-22.) The 360/50 was developed in Poughkeepsi^ 
ancT manufactured (assembled) in Poughkeepsie, Essonnes, and Mont- 
pellier. (Dunlop, Tr. 9 3649.) The 360/20 was developed in Boeblin- 
gen, Germany and manufactured (assembled) in Sindelf ingen, Vimercate, 
Italy, San Jose, and Boca Ratan. (Id.; Hughes, Tr. 71942-43.) System 
360' s SLT circuit packaging was designed in Endicott and East Fishkii:., 
and manufacturered in East Fishkill, Endicott, Essones and Sindel- 
f ingen. (Dunlop, Tr. 9 3649-50.) The 2401 tape subsystem was dev- 
eloped in Poughkeepsie, and manufactured (assembled) in Poughkeepsie, 
Essones, Montpellier and Boulder. (Dunlop, Tr. 



-341- 



13 
14 
15 
16 



23 

24 
25 






7 
3 



1 ! 



Within IBM, it was recognized that achievement of SPREAD'S 
recommendations would require "great effort" to "control and 
coordinate the work of several divisions and that of the IBM World Trade 
•Corporation". (Knaplund, Tr. 90470-71.) At the time of SPREAD there 
were 15-20 engineering groups generating processor products in IBM. 
(DX 1404A, p. 7 (App.. A to JX 38).) These groups resided in four 
principal areas — DSD, GPD, FSD (Federal Systems Division) and WTC (World 
Trade Corporation). (DX 1404A, p. 49 (App. A to JX 38).) If a single 
compatible line of processors was to be achieved, design control had to 
be centralized in a single location.* Accordingly, the SPREAD Committee 
recommended the establishment of a systems architecture group that would 

be charged with formalizing the design objectives for NPL and providing 

it 

illogical specifications for the hardware and software. (DX 1404A, p. 49 

• (App. A to JX 38).) Such a group — the NPL Architecture Committee — was 

formed in early 1962, and served in the role of "advisor" to the various 



A 



.0 

.7 ■ 93650.) The 1403N1 printer was developed in Endicott and manufactured 
•in Endicott, Raleigh, Sindelfingen and Vallingby, Sweden. (Dunlop, 

p S ;Tr. 93650-51.) The 2311 was developed in San Jose and manufactured in 
uSan Jose and Sindelfingen. (Dunlop, Tr. 93651.) The 2671 paper tape 

[g 'recorder was developed in LaGaude, France, and manufactured in Essonnes 
:jand Montpellier. (Dunlop, Tr. 93651.) 

:J * Centralized control of worldwide development efforts made good 
2i [sense from another standpoint as well. The SPREAD Report projected 

;;a very rapid increase in the growth of computer usage outside the United 
22 i States during the 1960s; whereas the average domestic growth rate was 

(projected to be 15%, the foreign rate was projected at 37%. (DX 1404A, 
?- ! |p. 11 (App. A to JX 38).) So large an element of demand obviously 
""" jcould not be ignored in the development of new products, and the 
2 4 ijCommittee recommended that the needs of users worldwide be taken into 

(account in all phases of NPL development. (DX 1404A, p. 49 (App. A 
„- : |to JX 38) .) 

£■3 i 



J -342- 

! 



L 
2 
2 
4-1 

f. 

7. 

8j 

5 i 

io i 

a! 

i 

12. ' 

is : 

i 

L4.i 

^ i 



NPL engineering groups. (Case, Tr. 74487-88, 74492-9 .) They held 
"dozens if not a hundred or more meetings" relating to NPL. (Case, Tr. 
74469.) 

On the. manufacturing side, too, a number of disciplines were 
: imposed to assure that there were no major discrepancies among the 
products produced on either side of the Atlantic. IBM's plants 
worked "very closely" together to develop "worldwide manufacture 
plans" and employee training plans. (Dunlop, Tr. 93651-52.) IBM also 
introduced, for the first time with System/360, the concept of 
"single engineering control". (Dunlop, Tr. 93641, 93646.) Under this 
concept any laboratory responsible for designing a part, component or 
product was also responsible for releasing that design to all the 
plants, worldwide, that were going to manufacture that part, component 
,or product. (Dunlop, Tr. 93641.) By introducing this concept, IBM 
was able to: 

(a) achieve a "high level of confidence" that all parts, 



1i» 
C if 

■| wherever in the world produced, would perform in a comparable 

17 \ 

,'| fashion; 

IS 

\ (b) achieve the ability to exchange parts or assemblies 



IS' : 

•J or products among manufacturing locations in times of tech- 

2G ] 

j nological difficulty or great demand; 

21- If • 

i (c) avoid duplication of engineering effort, since there 

j was no need to design the same product or component twice in 

j two different places. (Dunlop, Tt. 93642-43, 93645.) 

24 ! 

j Apart from the need to impose new disciplines, it was 

25 j " ■ 



-343- 



3- 
<• 

ri 
a 

.a 

■21 
14- i 



apparent that a "substantial" segment of IBM's "new product develop- 
ment resources in the electronic data processing (EDP) area" 
would be required to announce the New Product Line in the first quarter 
^|of 1964. (Knaplund, Tr. 90471.) Brooks testified that the original 
estimate for 360 programming was between $100 and $200 million. 
(Tr. 22706.) That estimate was exceeded by better than $25 million. 
(Id. ) Brooks' staff in DSD alone grew from "20 or 30" in June 1961 
to "several hundred" by February 1964. (Brooks, Tr. 22669.) A presenta- 
tion made to IBM Chairman, T. J. Watson, Jr., in November 1964 showed 
that IBM's annual research and development expenditures rose from 
approximately $175 million per year in 1961 to $275 million per year in 
1964. (PX 6671, p. 6.)* 

More investment still was needed to meet the requirements 
for SLT components. The 1961 decision to manufacture SLT in-house 
required a rapid buildup in manufacturing facilities and resources . 

(Knaplund, Tr. 90546; E. Bloch, Tr. 91562.) To meet the projected volumes 

» i 

lifor 360, IBM had to become "in a very short time, the largest component. 

| manufacturer in the world". (PX 1900, p. 9.) In 1961, IBM established 

;ja Components Division to "focus all of its resources in terms of both 

; manufacturing and development on that goal of making SLT components." 



'6 

17 ; 
IS 

19 

*:j 

.1 

21 ;i 

: - * It is interesting to note that, in the 1959-64 period, IBM's 

2- ^research and development (R&D) expenses were not only absolutely highe: 

;jthan some of its major competitors (Burroughs, NCR, Sperry Rand and 

-- (CDC) , but were more than double the expenditures as a percentage of 

^ jrevenue for Burroughs or NCR or Sperry Rand. Each of their ratios 

j 4 :jof R&D to revenue remained about level over that period. Among the 

"""" jfour, only CDC, which was developing the highly successful 66 00, 

0jC -(showed an increasing R&D to revenue ratio. (PX 66 71, po. 5-6.) 

I 

I -344- 



Li 
2 ; 

i 
i 

3 

i 

5 
c 
7 

a 

9 

LQ 

II 
12. 
13 

15 
15 
17 
IS 



(E. Bloch, Tr. 91562, 91891-92.) In 1963, the Components Division 
opened a new plant in East Fishkill, New York as a manufacturing 
development site for System/360 components. (E. Bloch, Tr. 91563, 
. 91891-92.) Prior to the 360 announcement, IBM hired "a large number of 
people" and "started to build additional buildings" in order to meet the 
anticipated SLT requirements. In addition, IBM's Endicott location was 
enlarged to help produce packages for mounting SLT modules (E. Bloch, 
Tr. 91892) and part of a plant in Essonnes, France, was converted into a 
"component facility" to help meet "worldwide requirementC s ] " . (E. Bloch, 
Tr. 91893, 91563-64.) 

Perhaps as significant as the magnitude of IBM's investment 
in 360 was the fact that all of those resources were being put into a 
single project: IBM was "putting a lot of eggs in one basket ... 
and the success of the company was in many ways to be determined by 
the success of that one project". (Case, Tr. 73561; Evans, Tr. 
101128.) If 360 were rejected by customers there would be few alterna- 
tives around for IBM to offer and none that was thoroughly funded or 
'covered a very large part of the product line. (Case, Tr. 73 562.) 



IjThus, once the die had been cast and the decision made to go forward 



19 ;• ■ , 
•;with the SPREAD Committee's recommendations, IBM's fortunes became 

:i " inextricably tied up with the NPL project". (Case, Tr. 73562.) Within \ 

2i i • 

:IBM and without, the 360 project came to be known as the "you bet \ 

22 "• ! 

'your company" venture. (Friedman, Tr. 50378; Case, Tr. 73561-62; j 



| Evans, Tr. 101126.) If that venture had failed, IBM would have 

2.L 'I 

jbecome a "radically different company, if even in tne computer 

~ :|business". (Evans, Tr. 101128.) 



-345- 



3 

.Q 



Despite the risk, IBM decided to develop the 3 60 line 
because " [Rethought that the System/360 development was the best way 
to more rapidly grow the market, more rapidly expand demand for our 
products". (Case, Tr. 73606.) It was the sort of risk that IBM was 
forced to take by competition if it was to succeeds Seemingly safer 
alternatives to 360 continued to be advanced within IBM right up to 
j ,j the time that 360 was announced, (See, e.g., Case Tr. 73589-92; PX 1074; 
PX 1090.) As they had rejected the 8000 series, IBM management rejected 
those alternatives because they would not have given IBM the kind of 
long range solutions that it needed in the competitive environment of 
the day. (Evans, Tr. 101277; see also DX 4806.) 

System/360 represented a price/performance improvement over 

:i 
.^ \ IBM's existing eauipment which Learson described as "a price reduction 

^ :! 

of 30-50%." (DX 1525.) Within IBM, it was recognized that no "single 
announcement" had ever "obsoleted so much existing equipment at one 
time". (PX 1099A.) IBM was forced to make such an announcement. 
[The SPREAD Committee had set as an objective the creation of a plan 
that would "optimize the conflicting demands" of "market need" on the 

is ;| 

lione hand and "impact on present installed processors" on the other 

® il 

: (DX 1404A, p. 7 (App. A to JX 38)) — but IBM had to impact its own line 

2G 

:jor stand by and watch others do so. 

21 -\ . 

;j In an effort to blunt the impact of System/360 on IBM's 

22. ; i 

{existing product line, IBM Treasurer K. N. Davis recommended that 36 be 

23 

{offered for sale only. Davis made the suggestion because technology and 

Z± \ 

jprice/performance were "changing and improving so rapidly" that 



-?;, 



-346- 



he believed it might be in IBM's interest to transfer to customers 
some of the risk of technological obsolescence. In addition, System/ 
360' s price/performance on a rental basis was so superior to existing 
. IBM systems on rent that customers would rapidly displace those 
systems with 360s. (Knaplund, Tr. 90511-12.) The recommendation 
was rejected because "IBM had to continue to offer a rental option 
in order to remain competitive": competitors offered that option and 
customers found it desirable. (Knaplund, Tr. 90512-13.) In this 
respect, IBM's experience was no different from its competitors. For 
example, McCollister testified that RCA offered its systems on a lease 
basis because the customers insisted upon it and because all other 
manufacturers in the industry offered it. (McCollister, Tr. 9292-300; 
see also Palevsky, Tr. 3145-46; Spangle, Tr. 5531; Oelman, Tr. 6160.) 
Indeed, customers as well as IBM could perceive that technology was 
changing and would not have been willing to accept the risk of obsoles- 
cence. Competition ensured that they did not have to do so. 

As Withington agreed, IBM had to introduce a product line 
comparable in performance and function to System/360 if it wanted to 
stay in business because its existing line would have become "obsolete" 

and unmarketable. (Tr. 56524, 56539.) Thus, IBM Vice President and 

2G I 

•I Group Executive Learson wrote to C. J. 3ashe, Manager of Technical 

i 

21 \ 

•Development, GPD, and T. C. Papes, Manager of Systems Development, GPD, 

*— \ in July 196 3: 



L 
2 

3 

4- 

£ 






7 

3 

a 

10 i 

! 

11 ! 

12 i 

I 

13 ; 
i 

14*; 

i 

15 ; 
i 

15 ; 

17 ; 

IS 

19 



24 I 



?«; i 



"The 101 [announced as the System/360 Model 30] must be 
engineered and planned to impact solidly the 14 01. 

"I know your reluctance to do this, but corporate policy 
is that you do it. It is obvious that in 1967 the 1401 will 



-347- 



be as dead as a Dodo bird.E*] Let's stop fighting this." 
(DX 1406.) 

Hughes testified that this letter was passed down to him 

i through the management chain to emphasize the importance of the 36 0/ 



3 

I 30 program and the company's policy with respect to that program. He 

; understood that the 360/30 would make the 1400 family obsolete — and 
5 ; 

! had to do so. (Tr. 33965-66, 33972-73.) Despite the fact that by 
7 | 1964 IBM had shipped thousands of 14 01 svstems, of which 75-80 

7 !l 

percent were still owned by IBM and on lease to customers, it was pert 
ceived that " [i]f we didn't obsolete it and replace it, someone else 
would". (Hughes, Tr. 33962-63, 33965; see also Tr. 34062.) That 
same view was echoed in a letter written by a Staff Vice President to 
the President of Southern Railway in April 1964, recommending the 
acquisition of 360/30s to replace Southern's 140^ s: 



8 

9 
10 
11 
12 
13 
14 
15 

16 ! 

i 

17 ! 

18 ! 
19 
20 
21 
22 
23 
24 
25 



"This will reduce the IBM rentals by $4,000 a month in 
Atlanta. There is also a good possibility that we will be 
able to eliminate the 1401 computer in Washington, using 
computers in Atlanta by tape to tape control from Washington. 
This would also save us $4,000 to $5,000 per month rental in 
Washington. Prices of computers have been coming down while 
the computer capacities are being increased tremendously. 
If IBM does not bring out new computers at reduced prices, 
their competitors take the business." 

According to John Jones of Southern Railway who helped draft that 

letter, it reflected his view of competition in 1964 — i.e., that if 

IBM and others did not bring out new products to meet competition, com-' 

petitors would take their business away — the kind of competition which 



*By year-end 1966, I3M had installed over 10,000 1401s, far and 
away the largest number of any system type that IBM had ever shipped 
at that tine. (?X 190 0, p. 7.1 



-348- 



L 
2 
3 

4*1 

5 I 

I 
/«■ ' 

r\ 

8 

9 

1Q 
U 
IZ 

12 



had increased "tremendously" since then. (J. Jones, Tr. 78991-97.) 

It was a view that was shared by IBM's competitors as well 

"There is no looking backward in our industry [the com- 
puter business] as you undoubtedly know. If one stops to 
ponder the past and be self-satisfied, the more aggressive 
competitors will quickly charge past." (Hindle (DEC), Tr. 7447 
DX 517, p. 2j 



and 



"It was our finding that the life of a family of com- 
puters was quite limited . . . and that you did not bring 
out a family of products that simply met the price/perform- 
ance characteristics of the then existing competition. You 
had to bring out something that would exceed the price/ 
performance of the existing competition because you knew 
full well that they were going to be moving ahead of you. 
It is a constant leap frogging game." (R. Jones (GE) , Tr. 8867) 



One gets "to a point in which the price/performance is 
so improved over equipment of days of yore that it is 
clear that . . . users are going to move to new equipment, 
and either [one is] going to provide that new equipment 
or [one's] competitors are going to provide it". (R. Bloch 
(Honeywell/GE) ,Tr. 7761-62; see also Hindle (DEC), Tr. 7448; 
R. Jones (GE) , Tr. 8865; Hangen (NCR), Tr. 10423-24, 10431; 
Currie (XDS) Tr. 15175-76; Brooks, Tr. 22705, 22795-96; Withincton, 
IS I! Tr. 56560, 56565; DX 426, pp. 7-8.) 

17 ; As we have already discussed, it was the recognition that 

IS ; competitors would supplant IBM's installed base if IBM took no action, 

19 i!as reflected in the SPREAD Report's "product survival charts", which j 

:j I 

2G ihad triggered the NPL project to begin with. The SPREAD Committee's j 
*" 1 I 

21 [prediction that IBM's highly successful second generation line would be j 

22 "[superseded by competition starting in about 1965, turned out to be | 

i ! 

23 ''accurate as to substance, but overly optimistic as to time. As j 
! 

24 'jWithington testified, the industry was in a state of "technological fer- ! 

I ' 

2= "ment during the period 1956 through 1964", with "new technologies . . . \ 

■i • 

j -349- 

i i 

'i I 

! ! 

,t I 



/ 
3 
3 

a 

2 
3 



new types of components . . . [and] significant software products . . . 
being invented and employed at a rapid rate" and new models of computer 
systems superseding older computer systems at a "rapid rate" and 
achieving "relatively rapid success in the marketplace". (Tr. 56459-60.; 
In 1963 and early 1964, the "leapfrogging" which was "characteristic" 
of the computer industry (R. Jones, Tr. 8846) had occurred. In July 
of 1963 Learson could say that "in 1967 the 1401 will be as dead as a 
Dodo bird" (DX 1406) because it was already being surpassed by newer 
models of computer systems. 

Indeed, at the highest level within IBM there was concern 
that the System/360 might not be enough of an improvement to recover 
its costs. Thus Watson, writing to Learson in June 1963, stated 
concerning the New Product Line: 



"I think it important to note, however, since we seem 
to have suffered for a few months or even years because our 
machines predated the effective competitive machines now in 
the marketplace, that we now make these [System/360] machines 
2 ij good enough so they will not be just equal to competition, 
for I am sure that once they are announced our competitors 
will immediately try to better them. This is all to the good 
and I am for competition, but I want our new line to last 
long enough so we do not go into the red." (DX 4806.) 



,a i 

.7 

.3 



i 

{Similarly, writing in November 1963 to a group of IBM executives, 

.} Watson said: 

•0 .1 
•I 



?7 ! 



"There is a great deal of running about and extra effort 
being expended in all areas of the IBM company now because 
-I once again we have allowed ourselves to become somewhat non- 
12 '] competitive without recognizing one simple obvious fact. In 

,| bringing new machines and devices to the marketplace, our 

23 j competitors in today's market are simply not going to stand 

j still. We should recognize that in every area, they will 

Z4. '; take the best we have and immediately start working in a 

j tough, hard-minded fashion to produce something better. 

* s ' 
> •■ 

I "We find ourselves in our present position because we 

! seem to assume our competitors will stand still in certain 



-350- 



L 
2 
3 
4- 

5 If 



a > 



7 1 

3 

9 

ia 

ii 

12 

13 
14- 
15 
La 
17 



areas after we announce a superior product .... 

"I believe that whenever we make a new machine 
announcement, we should set up a future date at which point 
we can reasonably assume that a competitor's article of 
greater capability will be announced. We should then 
target our own development program to produce a better 
machine on or before that date*" (PX 1077, pp. 1-2.) 

Charts prepared by DSD Market Evaluation Manager, J. C. 
Wick, comparing the price/performance of the New Product Line to com- 
petitive products in February 1964, showed that 360' s price/performance 
was superior to that of recently announced machines from RCA, 
Burroughs, CDC, Honeywell, Univac and GE, but also showed quite clearly 
that those competitive machines had a price/performance advantage over 
the earlier announced IBM machines of the 1400 and 7000 series. (PX 
1099A, pp. R2-R3.) We discuss some of the competitive announcements 
which created this situation in the histories of these competitors 
during the early 1960s. However, some of the announcements merit 
particular attention here. 

In October 1963, DSD President G. F. Kennard wrote to T. J. 
Watson, Jr., and A. L. Williams: "RCA has recently announced the 



.- =3301. . . . Initial performance specifications indicate that the 3301 
- S 1 

ilhas about 50 percent better processing capabilities than the IBM 7010" 

2- I at a comparable price. (PX 2952.) In November, 1963, it was 

i 
7t 'reported within IBM that GE was discussing in public a new series of 

^ machines planned for announcement before the end of the year. "In one 

•[case GE stated, system cost would be approximately the same as the 
22 I 

IBM 1410 but would be 40% faster." (PX 36 24, d. 4.) GE announced the 
24 i 

J 

,i 

•?« '< 

-^ i 

! 

•j -351- 

■i 

:i 



400 Series in December 1963,* and at the same press conference revealed 
the future availability of its 600 family.** (Weil, Tr. 7181; DX 488; 
DX 490..) The 400 series offered a 1401 simulator which permitted IBM 
1401 programs to be run on or converted "easily" to the 400. It was 
aimed at 1401 users* (Weil, Tr. 7031-34.) 
'r'l The CDC 6600, which CDC began discussing with customers 

i| 

l jj before announcement (Norris, Tr. 5937-38) in 1962 (JX 10, u 4), caused 
IBM Chairman Watson to ask "why we have lost our industry leadership 
position by letting someone else offer the world's most powerful 
computer". (PX 1045.) CDC's 3600, which had been announced in May 
1962, was viewed within IBM as "technically superior to the 7094". 
(PX 1026A.) By April 1963, 0. M. Scott, IBM Vice President and Group 
Executive, was reporting to Watson and others that "3600-type com- 
petition" was creating a "serious situation" and that such competition 
(from CDC's 3600 and 6600 and from Philco ' s 212) was able to offer 



7 ! 
S 

.9 

'a 



I "one-and-a-half to two times the performance of the 7094 at a lower 

price". (PX 1025.) Scott added that the 501 (360/70), as planned, 

would enable I3M "to favorably compete with the CDC 3600". ( Id. ) On 

April 23, 1963, Watson determined to "just sit tight" and stay with 

the 501 approach "unless the roof falls in", but wrote that IBM had an 
:u :j 

, active program in DSD called the "7094 B crime"' which vras sufficiently 
LI : ! 

^advanced to be announced in June 1963. (PX 2807.) Within two weeks, 
•2 ;, i 



?i | * GE announced the Models 425, 435, 455 and 465. The 455 and 465 
iwere never delivered. (Weil, Tr. 7181; DX 4 90, pp. 1-3.) 

15 I 

| ** The 600 family was actually announced in the summer of 1964 

'■'(Weil, Tr. 7197-98; DX 491, p. 1) and was aimed at IBM 7090 and 7094 
lusers. (Weil, Tr. 7033-38.) 
: ! 

1 -352- 



3 

4- 
5 



^ 

a 



cascading losses to CDC's 3600 caused a reevaluation of that decision, 
and Watson asked Scott to advise him when the situtation got "out of 
control". (PX 3619.) One week later, Scott reported back that IBM 
was repeatedly "being beaten" by CDC's 3600, 6600 and 1604, Philco's 
212 and Remington ' Rand ' s 1107. He recommended announcement of the 
7094-B' "at the earliest possible date". (PX 3620.) IBM announced 



^ .{the 7094 Mod. II on May 16, 1963 (DX 13958 ) r but this extension of the 
7090 series still "could not meet either the performance level or the 

price of a comparable CDC 3600." (PX 320, p. 15.) As a result, CDC's 
success with the 3600 continued unabated. (PX 320, p. 15.) With 
virtually all of IBM's development resources tied up on 360, IBM was 
simply unable to respond effectively at that time — all of IBM's eggs 
were indeed in the 360 basket.* (See Case, Tr. 73589, 73561; Evans, 
•|Tr. 101128.) In the meantime, CDC was able to achieve success "by 
concentrating on an area of IBM price weakness, and by showing a major 
price performance advantage to potential customers". (P.X 320, p. 15.)** 



3 ' 

* I 

j 

ii ! 

12. 



L4- : 



Perhaps most important of all, however, was the announcement 



IS , 

17 ; 
is ; 

xs i 

:■{ * At just about this same time CDC's chief development engineer 
?g '-for the 6000 Series, Vice President Seymour Cray, at CDC's June 1963 
'corporate planning meeting, urged that CDC announce the 6600 and a 



2i ^successor in order to "slug" IBM because he speculated that IBM had 
'; "rradea mistake in putting all [its] eggs in an integrated circuit 
kef. (DX 13526, Forrest, pp. 748-50O 

** No competitor was able to offer such an advantage once 360 was 



^ (announced. (PX 320, pp. 4-14.) 






£Z 



-353- 



J i 

7: 

3 

9 

a 

i 

2. 



of the Honeywell 200 in early December 1963. (McCollister, Tr. 11367; 

PX 1Q79; DX 167) This machine offered substantially improved price/ 

performance over the 1401. (McCollister, Tr. 11237; Knaplund, 

Tr. 90475; Evans, Tr. 101188; DX 167.) It also offered a conversion 

« 

program called the "LIBERATOR" which made the H-200 to a considerable 
degree compatible with IBM's 1401. (R. Bloch, Tr. 7605-06; McCollister, 
Tr. 11237; Goetz, Tr. 17652; DX 167; DX 488.) 

Within IBM the H-200 announcement was viewed as "even more 
difficult than we anticipated". (PX 1079.) Within two days of the 
announcement, Lear son wrote to T. J. Watson and A. L. Williams that 
the 101 (360/30) would have to be announced "as soon as possible"* and 
priced at its "lowest projection" in order to be competitive. ( Id. ) 
IBM's marketing force regarded the H-200 as a real challenge (Evans, 
Tr. 101186) and at least one person in I3M called it "the most severe 
threat to IBM in our history". (PX 3912.) By February of 1964, the 
Sales Division was "reeling from losses" to the Honeywell 200 and 
jj "wanting a more competitive answer". (Evans, Tr. 101196.) 3ecause of 



lithe H-200, IBM's Data Processing Division continued and intensified 

LS \ 

;jits pressure for the earliest possible announcement of System/360, 

19 i 

'•{earlier still than even the then-planned mid-March announcement date. 

IQ i 

:j (Knaplund, Tr. 90475; JX 38, 11 16; PX 1095; DX 2983.) 

21 :! 

1 As competitive pressure mounted, the debate whether to go 

; j forward with 360 as planned or to announce extensions to the existing 

I 

"" j * The target announcement date at that point in time was March 196 4 
- s ) (PX 1079.) 



-354- 



L 
1 
3 

4- 

a 

m 

a 



product lines was rekindled. The latter approach would be safer and 
easier: it would not be as "revolutionary" as 360 and would therefore 
run a lower risk of user rejection. (See, e.g., Case, Tr. 73590, 73512; 
Evans, Tr. 101127.) Moreover, it would not require users to convert 
their existing applications programs. In November 1963, IBM's 
Corporate Staff advanced the position that "new marketing developments"* 



the announcement, in May June 1964 of "several improved current line 
systems — such as the 7074X, 7010X and 7094X" . In their scheme of 
things the NPL announcement was to be put off for 6 to 12 months. 
(PX 1074, pp. 2-3.) 

The Honeywell 200 announcement provided perhaps the sharpest 
temptation to depart from the System/360 plan. In early 1963, IBM had 
?<t ;j a 1401 built out of SLT circuitry to establish the feasibility of using 
I SLT in the New Processor Line, ("Hughes, Tr. 33952—53.7 McCarter, 
I Tr. 88394; JX 38, 1( 7; DX 4800.) The Honeywell 200 prompted 
I sharp debate within IBM whether a new technology (SLT) version of the 
| 1401 (called the 140 IS) should be brought out and the 360/30 announce- 
ment delayed or cancelled. (Hughes, Tr. 33953-54; Evans, Tr. 101188, 



is 






17 
IS 



13 

;| 101195.) The chief proponent of this new plan was GPD President John j 

'* i 

IHaanstra, who had been Chairman of the SPREAD Committee. Haanstra 

21 1 . | 

il 
22 :; 

<?- ; ! 

- 3 j * Thsse new developments included the announcement of competitive 
| j processors offering easy conversion to IBM customers and other new com- 

*- jpetitive offerings with improved price/performance as well as "the 
;« continuing unattractiveness of programming conversion and associated 

*5 ..expense to our customers". (PX 10 74, p. 2.) 

1 ! 

;j -355- i 

I ! 



2 

3 

3 
C 

7 
3 
9 
.0 

12 
13 

14 



believed that the 1401 was a "fundamentally sound" approach to meeting 

user needs and that the 360/30 approach was "improper" because it 

created the exposure of requiring customers to convert: 

"We must have a position which sticks to the 1401 as a 
fundamentally sound andproper [sic] method for commercial data 

processing. I do not believe that we should in the GP small 
machine area imply in any fashion whatsoever that the 1401 
approach to problem solving is out of date and that people 
must change. 

« • » • 

"... [I]n the final analysis we must sustain a position 
of 1401 as a right programming approach now and into the 
future. An approach which implies that we must convert is 
basically improper." (PX 3913.) 

The Data Processing Division, however, regarded the 1401S as 

only a fallback position in the event that the 360/30 was not ready 

soon enough or was not good enough: 



"The best solution ... is a 101-H machine with a 
competitive price to the H-200 and a performance equal to 
or greater than the H-200, ready for announcement by mid- 
February. . . . This system would not only compete head 
r g ; on with the H-200 but offer the customer the opportunity 
to grow in the NPL line, which is the direction we want 
them to take. 






j "The 140 IS machine, which has been discussed, is a 

i second choice to the system described above and has been 
! supported by us only because we have not received a 

commitment that the 101-H machine could achieve the per- 
formance desired or meet an early announcement schedule." 
(PX 1090.) 



~ ! Evans was sure that it was a mistake to produce the 140 IS instead of 
4 

^ | the 360/30, and that it would not make sense to do both. As earlv as 

r- l 

- 3 j September 1^63 he had inveighed against "continual competition with 

-- ! temporary machines" because they would "only dilute [IBM's] already 
25 | overcommitted resources and abilitv to meet the NPL challenge". (DX 

i 

1 -356- 



L 
2 

4- 



3 

a i 
7{ 
3 
9 

ia 

u 

12 

13 

15 

IS 



2983.) In his view, if the 1401S had proceeded, it would have "delayed 
if not killed" the 360/30 and "wreaked havoc with the costs of the rest 
of the System/360 line". (Evans, Tr. 101195-96.) In addition, Evans 
regarded a decision to produce the 1401s as relegating the NPL more to 
the scientific area and signalling "a discrete scientific line/ probably 
along the 7090 philosophy particularly if competition does the H 200 
type of thing to the 7090 family". He felt this would erode the basis 
for NPL and lead to a processor policy of "discrete 1400-type commercial, 
discrete 7090-type scientific, plus various custom units for new 
application areas", as "the inevitable conclusion". (Evans, Tr. 101275- 
76; PX 6668 (DX 14514).) As we shall see below (pp. 379-81), GE 
ijwas in fact attempting to do "the H 200 type of thing to the 7090 family. 
Evans was right . 

Although contingency plans were laid for a possible February 
1964 announcement (PX 6202) , IBM decided not to proceed with the 
1401S. Evans testified that the 1401S was ultimately rejected 



: "[b]ecause the evaluations and conclusions of senior management 

17 ; were that it was not an advanced system that would solve the 

■\ applications of the future as we then saw them — that ... it 

IS •! was a machine that would not have long life and would not be 

:j competitive for more than a short period, and that the 36 

13 j family plan with all of its advanced features and functions 

:( and capability and the unusual power it brought the users was 

2Q'! a substantially better plan". (Tr. 101277.) 

2i 'in short, the 360/30 was expected to be "a better overall performing 

22 \ system than the 1401 had been or could have been, had we extended its 

22 [life". (Hughes, Tr. 33953-54.) 

\ 

2* \ a. Preparation for Announcement. It was clear by the end 

.i 
-- lof 196 3 that announcement of Svstem/36 was required for IBM to remain 

i 

:1 

■! -357- 



L 
I 
I 

4. 



competitive. (Knaplund, Tr. 90475.) We have already discussed how, 
beginning in 1961, IBM began applying massive resources to the NPL 
project. Evans testified that the "whole 360 program had been on a 
. crash basis . . . since almost inception" and that by the latter part 

„ 

of 1963 it had become an enormous program with its own inertia". 



„ !' (Tr. 101190, 101198^99.) In December of 1963, development of the line 
was "on or ahead of the schedule called for two years earlier in the 
SPREAD report" (Knaplund, Tr. 90477),* and "two of the prime movers of 
the project, Evans and Brooks, were recommending announcement of the 
entire family in the first part of 1964.** 



/ 
3 

.0 

1 i 



* A PERT chart (DX 1405), prepared by Ernest Hughes in October 1962, 
laid out the job to be done in order to accomplish the Model 30 
program. (Hughes, Tr. 33933-34, 33947.) The chart showed that the 
Model 30 would be ready for first customer shipment on August 1, 1965, 
if the sequence of events identified on the chart were "successfully 

.4 j' completed". (DX 1405; Hughes, Tr. 33947.) According to Hughes, all 
of those tasks were completed "close" to the dates projected for their 
completion back in 1962, and the first 360/30 was actually shipped in 
June 1965. (Tr. 33947-49; see also JX 38, 1f 24.) This was so despite 

'£ !| the fact that IBM's Product Test organization was of the view that the 
'" System/360 Model 30 central processing unit was farther behind in their_ 
testing procedure than any of the other System/360 central processing 
units announced in April 1964. (PX 1107, p. 7.) 



.7 

S 

:j ** By the time 360 was announced, engineering models of all the 

: q ; | processors had been built (Brooks, Tr. 22695-96); full instruction set 
k ii compatibility across the five processors had been achieved (Brooks, 
?rr j.Tr. 22785) ; a complete processor had been built usinq SLT tech- 
" M -inology and demonstrated to establish the feasibilitv of the 
. 7 ;jnew circuitry (Hughes, Tr. 33952-55; JX 38, 1f 7, p. 5; DX 4800); 
~ 'jmany thousands of SLT modules had already been produced (DX 4796, p. 8) 
~ 1 most of the processors and some of the peripheral equipment were in 
— jthe early stage of product test (McCarter, Tr. 88383; JX 38, If 19); 
,- ilall, or almost all, the memories had undergone technical evaluation 
^ [testing (Brooks, Tr. 22699) ; microprogramming and multiprogramming had 
m4 Ibeen tested on the Model 40 (McCarter, Tr. 88382-83); and four esti- 
-" [mating, forecasting and pricing cycles had been completed (DX 1172, p. 

;2). Product Test had been involved with the development from the 
- 5 :j beginning (McCarter, Tr. 88375; DX 1165): by the time of announcement, 

I 

:[ 

'•! -353- 



L 
2 

5 
4. 

3" 

7 

3 

a 

La ; 

Li 

12 

13. 

14- 

L5 

IS 

17 

IS 

19 

2C 



In September of 1963, Evans wrote to DSD President Kennard: 

"NPL is good — it is simple and powerful — it is ready enough 

— proven enough. IBM should go forward with . . . full 

announcement in the first or second quarter of 1964 with . 
programming systems committed." (DX 2983.) 

Although the SPREAD Report had not recommended announcing the entire 

NPL family at once, by December 1963 it was plain that there were 

powerful reasons for doing so. On December 27, 1963, Evans proposed 

that the NPL family be announced as a group in March 1964: 

"[Tlhe customers must better understand the abilities 
of the architecture and conversions necessary. It would 
be unwise of us to announce systems sporadically in an 
effort to optimize market penetration or profit. It is 
proper that IBM announce all the systems in a group so 
that our customers have the benefit of the family and 
can properly plan." (DX 4815; see also Evans, Tr. 
101072-75.) 

Less than one month later, Brooks wrote to Gibson, Haanstra and 

Kennard, stating that the equipment was "technically ready for 

announcement" and recommending announcement on April 7. (DX 1172.) 

He emphasized that System/360 " must be announced at one time" (id., 

p. 3) : 

"Piecemeal announcement would utterly confuse and misguide 
the customer in his planning. He could not make the best 
selection from the available models until all the models 
are announced." ( Id. ; see also Knaplund, Tr. 90486-88; 
Brooks, Tr. 22782-84.) 



~ i! thousands of tests had been made and "literally hundreds of problems 
^ ;!and potential problems" had been identified and resolved. The compon- 
^— j entry, systems and product testing program already completed was more 
„_ i| extensive than the entire program IBM had previously undertaken for 
^ {any system. (McCarter, Tr. 88390-93; Evans, Tr. 101065-66, 101082; 
|DX 1172, pp. 2, 5; see also DX 4815.) 

| RCA, Honeywell and GE all announced systems that were, by com- 

- 5 Jparison, in an embryonic stage of development. (See Spangle, Tr. 

14997-99; Weil, Tr. 7232-35; McCollister, Tr. 9635-41.) 

1 
1 

■I -359- 



Knaplund testified that he "understood that simultaneous announce- 
ment . . . would place an unprecedented load on the development of 
manufacturing resources of the product divisions " • However, the 
•advantages outweighed the risks* Since IBM was unquestionably 
"going to produce a compatible line, only by fully informing 
customers as to the full compatible range, including prices and 
functional specifications, could they evaluate properly IBM's 
offering. "It was my business judgment that partial announcement 
by IBM would result in customer confusion, superseding orders 
following subsequent IBM announcements , and churning of the order 
backlog in IBM's production schedules." (Knaplund, Tr. 90486-88.) 
The March or April announcement dates recommended by 



,4 
5 
5 
7 ; 

,3 ; 

a 



m ! l Evans and Brooks were virtually mandated by the first shipment 
j| dates planned for the 360 processors, which ranged from June 
1965 for the 2030 to January 1966 for the 2070. (See JX 38, 
1 24.) It was "generally industry practice on most computer systems 
[at that time to announce a system at least a year, and frequently as 

:jmuch as two years, ahead of the actual first delivery".* (Weil, Tr. 

II 

;(7064; see McCollister, Tr. 9635, 9641, 9646; Hangen, Tr. 10761-62; 
3 i| 

;JKnaplund, Tr. 90483-84; PX 355, pp. 33-36; PX 2226A,pp. 13, 19, 27; 

JPX 2432, pp. 19, 22, 28; DX 573; DX 4769; DX 4774; DX 8962.) 



1 " * There were "practical reasons" for this procedure from both the 
23 'manufacturer's and the customer's viewpoint, each of whom needed time 

(to prepare for delivery and installation. (Weil, Tr. 7064-65; 
?A ;|Withington, Tr. 58738-46; J. Jones, Tr. 79034-36; Akers, Tr. 96537-40; 

•DX 3726.) 

^ 1 

t 

i 

I • -360- 

i 

i ' 

i 

! 

■\ 
•l 



1* ' Such lead time was particularly important in the case of 

2. || System/360. Thus, Brooks wrote in January 1964: "The breadth of 
System/360 and the number of innovations, particularly in gross 
systems concept, will require substantial lead time between announce- 
ment and proper installation." (DX 1172, p. 1; see also DX 3726; 
DX 4815.) That time would be necessary to: 

(1) permit customers to replan their applications and take 
advantage of 360* s new concepts such as file orientation, 
communications facilities and large memories ; 

(2) permit customers to assimilate the "sheer amount- of new 
abilities, new options, new specifications, and new prices" 
that 360 would provide and select the best configuration of 
equipment to perform their applications; 

(3) permit IBM and customers to educate their personnel and 
prepare them for proper installation and maintenance 
of 360; 

(4) permit IBM to avoid deferred installations and conse- 
quential inventory build-ups; 

;j (5) permit customers to determine the need for and submit 

19 :i- 

;[ 

;i RPQs for special requirements; and 

20 J 

:i (6) permit customers to prepare their physical sites for 360 
22. -j - 

j installation. 

(Withington, Tr. 58738-46; J. Jones, Tr. 79034-36; Knaplund, Tr. 90483- 

88; Akers, Tr. 96537-41; DX 1172, pp. 1-2; DX 3726; DX 4815.) As 
24. 

'Southern Railway's President was advised by his EDP staff in 1964, 
25 j 

1 -361- 



3: 

5 
Si 
7 i 

3 

a 

10 

11 

12. ! 
13- 

!A !. 

15 

lo 
17 
IS 



r ; 

m 

1 | 

i 

I 
t 

J I 

z 
7 j 

3 j 

i 
a | 

I 

a I 

i 

i \ 

2 • 

^» 

J 



,6 
.5 



"there is always a year to 18 month delivery' lag from ordering to 
delivery. This amount of time is usually . . . required for planning 
and programming" . (DX 3726.) 

Over the course of the NPL development, there were numerous 
proposed announcement dates considered by various IBM personnel, 
ranging from mid-1963 to mid-1965. (Brooks, Tr. 22796; JX 38, 1( 15, 
p. 8; PX 1079? PX 1092; DX 1404A, pp. 57, 70, 119 (App. A to JX 38); 
DX 4782; DX 4786; DX 4790; DX 4814; DX 4815.) In December 1963, Paul 
Knaplund was assigned responsibility for assembling the technical 
evaluations, forecasts, cost analyses and profit projections that IBM 
top management would need to address the 360 announcement decision. 
Beginning in January 1964, he conducted weekly meetings with IBM line 

S and staff management to identify and assess the magnitude of outstand- 
ing problems and outline programs to solve those problems , so that he 
and they would be prepared to make judgments and advise top management 
on the advisability of proceeding with the 360 announcement. (Knaplund, 

i|Tr. 90474-77.) 

On March 18, 1964, IBM Chairman T. J. Watson, Jr. made the 



!j final decision to announce all of the models of the new line simul- 
:ltaneously on April 7, 1964. IBM's Product Test Department did not 

isupport the April 7 announcement — all other departments whose effort 

II ' 

'was required to provide the products, features and services offered in 

?*? 

~ ; jthe System/360 announcement did support it. (Gibson, Tr. 22648; 

"* JBrookG, Tr. 22799-800; Hughes, Tr. 34003; Knaplund, Tr. 90483, 90493; 

.1 
Zt jE. Bloch, Tr. 93311; JX 38, If If 18, 22; DX 1165; DX 9161.) 

■I 

I -362- 

•i 
i 

i 



z 

3 

4- 

f 
5 
7 
3 
9 

ia 
n 

12 

13 
14- 
15 



Evans, Brooks and McCarter explained the organization and 

role of the Product Test Department, in addition to all the other 

testing that was done at the time within IBM; how IBM management 

used the Product Test position to isolate problems and challenge 

the product development personnel to determine how they would solve 

those problems; and how Product Test, after it took its non-support 

position, later supported the shipment of System/360 to IBM's 

customers. (McCarter, Tr. 22568-70, 88362-93, 88434-55; Brooks, Tr. 

22786-88, 22850-53; Evans, Tr. 101065-66, 101083-95, 101174-78; 

PX 2126, pp. 2-5, 35-37; PX 4005; DX 1165; DX 1172, pp. 2, 5; DX 1409; 

DX 4815; DX 8083.) As G. B. McCarter* testified, 

"It did not follow from Product Test's non-support of 
March 16, 1964, that IBM could not or would not deliver what 
it committed to customers. ... To the contrary, Product 
Test's input was one of the mechanisms, like internal targets, 
designed to ensure that it would." (Tr. 88404.)** 



In fact, the processors announced on April 7, 1964, were 
-- ji all shipped on or before the dates estimated for shipment at 

announcement, except that the 2060 and 2062 on the one hand, and 
the 2070 on the other, were superseded by faster memory versions 



17 | 

IS \ 

I 

19 ! 



22 ,| * McCarter was DSD Manager of Product Test, and was the person who | 
j presented the position of the Product Test organization for all IBM • J 

21 ;j divisions to IBM management prior to System/360 announcement. (McCarter! 

•i Tr. 88373, 88380-81.) , j 

22;! j 

\ ** Prior to 360, there had been numerous occasions on which IBM j 

23 j announced products without Product Test support, including the 140 3 ; 
,! printer; 1302 disk file; the 709, 7090 and 7074 systems; and more than 

24- ! two dozen software programs. (McCarter, Tr. 38371-72, 88602-05; Evans, j 

I Tr. 101093-94; DX 4768; DX 7630; DX 9005.) j 



-363- 



1 called the 2065 and 2075, respectively, which were delivered on or 

2 | before the dates planned for their predecessor processors in April 

3 ; 1964. (JX 38, If 24.) Those first shipped systems, as planned, were 

4 • made available with the simpler operating systems offered with 360. 

5 ; (Brooks, Tr. 22853.) However, as we mentioned earlier, there were 

6 '• "significant schedule slippages in OS/360 software", (the most ad- 

7 vanced operating system for 360) which meant that some customers 

8 "received the full announced capabilities later than originally 

9 '. planned". (JX 38, II 25; DX 4740: Evans, Tr. (Telex) 3933-34; Welke, 

10 I Tr. 19410, 19631; see also Enfield, Tr. 20947-43; PX 4834, p. 23.) 

11 The problems with OS/360 occurred even though Product Test "cumula- 

12 tively did more testing of OS/360 than we ever had before for any 

13 set of programs for a particular system" (McCarter, Tr. 88390-93) , 

14 and despite the fact that IBM's programmers believed prior to April 

15 7, 1964 that they could produce OS/360 "in the way that it was ori- 

16 ginally intended". (McCarter, Tr. 88390-93; Evans, Tr. 101119; DX 

17 | 5609.) IBM, like the rest of the industry,* misjudaed the "enormous 

! 

18 i complexity" of developing complex operating systems. (Perlis, Tr. 

19 ]320, 2001-03; Spangle, Tr. 4997-99; Weil, Tr. 7215-21; McCollister, 

20 9696-98; Welke, Tr. 19281-82; Brooks, Tr. 22762-63; Withington, Tr. 

21 55914, 56729-30; McCarter, Tr. 88390-92; Evans, Tr. 101119.) 
22 



23 *Withington testified that "all manufacturers attempting the most 
advanced systems programs in that time had difficulties". (TR 56729- 

24 30.) For example: 

25 (a) Burroughs announced but never delivered the 38500 and 
B7500, in part because of software problems. (Perlis, 



-364- 



tj 

2 I 

i 

! 
5 ! 

I 

I 
S ; 



In a way, the modularity and standard interface of the 
System/360, which made hardware testing easier, as already discussed 



2 
W 

II i| 

i. 

13 



16 , 

17 [ 



Tr. 1320-21, 2001-03; Withington, Tr. 56599-600.) 



(b) Univac was compelled to delay the introduction of 
its EXEC 8 operating system for two to three 

years. (Perils, Tr. 2001-03.) Earlier, Lawrence 
7 jl Livermore Laboratory was compelled to rewrite "com- 

| pletely" the software that Univac had provided with 

3 i the LARC computer because the Laboratory was "not 

satisfied with it". (Fernbach, Tr. 517-18.) 



(c) Xerox "had difficulty producing the UTS [operating] 
system that [it] had announced". UTS was delayed 
for several years, costing XDS several millions of 
dollars in revenue. (Perils, Tr. 2001-03; Currie, Tr 
15303, 15352-54.) XDS also experienced delays 
in its XOS operating system. (Currie, Tr. 15704.) 



(d) The MULTICS operating system was never delivered 
by GE, even though GE, MIT and Bell Labs believed 
it could be feasibly designed. Honeywell finally 
completed the development three years behind the 
original schedule. (Weil, Tr. 7232-35; Wright, Tr. 
13373-76; Withington, Tr. 56730-31.) GE also 
had difficulty in making GECOS perform to their 
customers' satisfaction. Three different versions 
were eventually constructed, and none ever met the 
advertised capabilities. Because of those diffi- 
culties GE withdrew its Models 625 and 635 from 

IS :[ the market for a year or two in late 1966 or early 

:| 1967. (Weil, Tr. 7215-21; Withington, Tr. 

15 :j 56730-31.) 

22 .; (e) The Honeywell 8200 was unsuccessful, in part, 
.1 because of software development difficulties. 

21 ;| Honeywell had to spend "large amounts of money, 

i more than we had planned" to develop the soft- 

22.! ware. (Spangle, Tr. 4997-99.) 'Honeywell also 

;j took longer than anticipated to develop its 

23 j Series 60 line because of "difficulties in 
I developing software and microprogramming". 

24 1 (Spangle, Tr. 5008.) 

25 (f) RCA's TSOS was delayed "on the order of six to 



-365- 



(see above, pp. 360-62) , made software testing harder. It allowed 

customers great flexibility in the range of configurations which they 

could choose, and that, coupled with the wide variety of ways in which 

OS/360 could be used, led to "a very complex hardware-software system" 

which was literally impossible to test adequately. (McCarter, Tr. 

88544-45.) As Enfield testified: 

"Systems software by its nature cannot be adequately tested 
in a single environment but must in fact be tested . . . 
in a user environment in order to establish the many 
different types of configurations, the many different types 
of generation, options, the many different types of operat- 
ing environments. 



"If you were to take the various permutations of the 
options available to the user, the number of different 
tests that would have to be performed [in testing systems 
software] would exceed the time available for testing. 
I am talking about millions of different permutations and 
combinations of features that can be selected by the users . 
To test in each of those environments would preclude the 
issuance of first release of any operating system .... 
[bjecause as soon as you got around to testing the 999,000 
somebody would come out with another option and you'd have 
to go all the way through it again." (Tr. 20294-97; see also 
Perlis, Tr. 1347-48.) 

Only by expending "considerable internal efforts" was IEM 



j 

VI "i 



:j twelve months, possibly more", and performed poorly 

£ j and unreliably. (McCollister, Tr. 9694-95, 9707-08; 

| Rooney, Tr. 12132-34.) RCA's VMOS also experienced 

jt .; delays, which were estimated to constitute a "poten- 
tial problem" of some $2 million in monthly rentals from 
lost accounts plus a loss from delayed installations of 

~ j $3 1/2 million. (Rooney, Tr. 12335-36, 12349-50, 12358; 

T2 ! Conrad, Tr. 14088-89, 14133; DX 872, p. A.) The 

j difficulties with and instability of TSOS/VMOS 

7;t ; | "endangered [RCA's] position with any -customer who 

j had equipment on order and who planned to use this 

5S ! operating svstem" . (McCollister, Tr. 9704-05, 

~ -! 9710-11.) 



-366- 



L 
2:1 



able to remedy the problems with OS/360 — but IBM did so and provided 
customers a "very sophisticated, very complex software system, a 
software system that permitted the customer a great deal of flex- 
* | ibility . . . the customer could do a great deal with a minimum amount 
-'{of effort", which in turn caused System/360 "to show steadily in- 



creasing performance relative to competition and remain saleable 
longer". (Perlis, Tr. 1887-88; Palevsky, Tr. 3180; Rooney, Tr. 12576; 
Currie, Tr» 15186; Welke, Tr. 17308-13; McCarter, Tr. 88389; PX 1900, 
pp. 3-4, 8; PX 4833, p. 16; PX 4834, p. 23.) 

b. 360 's Success and Impact on IBM. System/360 was 
launched on April 7, 1964, and the internal doubts about its reception 
were soon dispelled* (See Knaplund, Tr. 90515; DX- 4740: Evans, Tr. (Telex) 
3932-33.) Orders for the systems "far exceeded IBM's forecasts" 
(Gibson, Tr. 22636-37; Case, Tr. 73258; Knaplund, Tr. 90547; Evans, 
Tr. 101123; Cary, Tr. 101780-81; JX 38, 1f 28; PX 1900, pp. 7, 10; 
DX 9331) and exceeded by thousands IBM's production plans which were 
Ibased on those forecasts : 



20 \ 



Gross Orders 
Estimated Actual Booked 



~ |l965 589 668 4,487 

1966 2,897 3,132 • 4,526 



22 1 



** 11965 & 1966 3,486 3,800 9,013 



_, ii (combined) 
24. i 



25 i 



-367- 



(JX 38, <f 28.) By October 1966, IBM's 360 order backlog represented 
an income of "almost three times . . . [IBM's then-current,] worldwide, 
annual sales of all products". (px 1900, p. 10.) 

As we discussed earlier, IBM management authorized sub- 
stantial increases in plant capacity prior to 360' s announcement in 
order to meet anticipated production and delivery requirements — 
including the establishment of an SLT manufacturing plant in East 
Fishkill, N.Y., and the addition of a new building at IBM's Endicott, 
N.Y., plant site for the manufacture of SLT cards and boards. (See 
above, pp. 344-45.) It was management's judgment that these manufac- 
turing capacity increases "adequately provided for the component 
and box production volumes required to support the System/360 
announcement together with planned future announcements". (Gibson, Tr. 
4JJ 22635-37; Knaplund, Tr. 90545-46; E. Bloch, Tr. 91895-96; DX 7691, 
-! p. 4; DX 9333.) However, because the total orders were far beyond 
~ \\ what was forecast and because larger size processors and more memory 



1 

a 



and peripherals than anticipated were being ordered, the demand for 
SLT modules also far exceeded IBM's expectations. (Knaplund, Tr. 



J 

,S :i 

,_ 1 90547; E. Bloch, Tr. 91899-906; Dunlop, Tr. 94774-75; DX 9331; DX j 

J :j ! 

;l 9332; DX 9333; DX 9334.) By May 1964, only a little more than one \ 

:Q -1 ! 

. i month after announcement, the projected "maximum annual ^fcdule Be- j 

d \ I 

■j quirements" had increased from 70-90 million to 130-190 million. (DX j 
•I 9331; see also E. Bloch, Tr. 91899-900, 91905-06; Dunlop, Tr. 94774- 

22 :i i 

i 75; DX 9332; DX 9333.) i 

2-i . ! 

• It was plain that the manufacturing capacity planned at j 

'j ; 

i 

■i ; 

1 

; | -368- 



L 
2 
2 

3 

s 

71 

3 

9 

10 ! 
li 

121 
13- 

14 



announcement would be insufficient and IBM began moving to meet the 
increased demand. By the third quarter of 1964 additional component 
production capacity was approved as an addition to IBM's Burlington, 
Vermont, plant site, and plans were initiated for additional assembly 
plant locations. By the end of 1964, IBM top management had approved 
expansion of the Federal Systems Division's Owego, N.Y., plant "to 
increase manufacturing capacity for SLT cards and boards"; and in 
the first part of 1965, two new plant sites in Boulder, Colorado, 
and Raleigh, North Carolina, were approved "to increase IBM's 
overall EDP manufacturing capacity". (Knaplund, Tr. 90547-48; E. 
Bloch, Tr. 91905-08; Dunlop, Tr. 93670; PX 5771, p. 28; DX 9038.) 
In addition, IBM provided special tools and training to Texas Instru- 
ments employees so that Texas Instruments might serve as an additonal 
source for SLT components. (E. Bloch, Tr. 91908.) 

By October 1965, IBM announced that it was "completing more 
than three million square feet of new manufacturing space" to meet 
requirements for System/360 — including plants in Boulder, Colorado; 
Raleigh, North Carolina; Montpellier, France; Vimercate, Italy; and 
expansions of existing facilities in Owego, Fishkill and Endicott, New 

iJYork; Burlington, Vermont; and San Jose, California. (DX 9038.) New 

20 1 

^plants were later added in Boca Raton, Florida and Brooklyn, New York. 



16 
17 

is 



21 j| 



j (Dunlop, Tr. 93670.) 



IBM also began hiring substantial numbers of new employees 

(Between year-end 1964 and year-end 196 7 IBM increased its work 

m * \ 

jforce by approximately 50% — adding more than 70,000 new employees. 

25 ; ■ 

! -369- 



(PX 5771, p. 3.; DX 13680, pp. 3-4; see also Knaplund, Tr. 90549-50; 

2 ! Dunlop, Tr. 93670; DX 4740: Evans, Tr. (Telex) 3934.) Evans testi- 

3 : fied that it was "an enormous job" to get the supply of parts flowing 

4 \ hire the people and train them in order to meet 360 commitments. 

5 ; At one point, IBM "even rented a circus tent to temporarily store 

6 • parts" until more permanent facilities could be secured. (Knaplund, 

7 Tr. 90549-50; DX 4740: Evans, Tr. (Telex) 3934.) 

8 In January 1965, IBM combined all product division manu- 

9 facturing functions in a single manufacturing division. It was 

10 ! believed that "by unifying responsibility for scheduling and produc- 

11 ing all the principal System 360 equipment, . . . manufacturing effi- 

12 ciency could be increased and information flow accelerated". 

13 (Knaplund, Tr. 90548-49.) The Systems Manufacturing Division (SMD) 

14 was thus created, with former GPD President C. E. Frizzell at its 

15 head. ( Id. ) By June 1965 Frizzell reported to IBM management that 

16 the production buildup would enable IBM to meet product shipments 

17 ! committed to customers. (Knaplund, Tr. 90550-51; DX 1154; DX 1155; 

18 J see also E. Bloch, Tr. 91915; DX 9333.) 

19 Within a few months, however, an "unforeseen" technical 

20 difficulty developed in the production of SLT technology. (Knaplund, 

21 Tr. 90551-52; E. Block, Tr. 91915-18.) The problem took about three 

22 months to sold, despite intensive efforts by IBM to do so, and the 

23 S delay put I3M several months behind the schedule for SLT production 
j 

24 j needed to satisfy existing customer commitments. (Knaplund, Tr. 
i 

25 I 90551-52; E. Block, Tr. 91917-19.) This was reported to I3M Chairman 



-370- 



L 
2. 
3 
4* 

5 
7 
3 

9 

10 

II 
12 

13 

14- I 



Watson , who immediately informed IBM's Board of Directors and issued 
a public statement advising that "during 1966 most System/360' s 
will be delivered 60 to 120 days later than originally scheduled" . 
(Knaplund, Tr» 90551-52; DX 9038.) Knaplund testified that, but 
for the unanticipated production problems, System/360 shipments at 
that point "would have continued on the committed plan". (Tr. 
90552-53.)* In the end, although many 360 hardware deliveries were 
made as scheduled and committed, there were some significant schedule 
slippages despite all of IBM's efforts to prevent them. (See 
Knaplund, Tr. 90849-54; JX 38, U 25.) 

The production, delivery and installation of System/360 
required a massive effort on IBM's part, which placed a severe 
strain on the corporation. (Cary, Tr. 101359-60; PX 1900, pp. 4, 8; 
DX 4740: Evans, Tr. (Telex) 3932-34; DX 8386, pp. 107-08, 111; DX 13677, 
p. 5; DX 13678, pp. 6-7.) In November 1965, Watson wrote to all 
I3M managers: "We're carrying out an assignment that in many 

„_ ! [ respects is one of the largest and most complex ever given to an 

;i 

I industrial electronics organization — almost a complete replacement 

of our principal product line". (DX 8886, p. 107.) It was a task 



IS 



18 i 



13 i 



„ f that some in IBM likened to "trying to swallow an elephant". 

20 ! 



m I (Carv, Tr. 101359.) 



J As we have discussed, IBM had to build new facilities and 

22 . 

23 I 

| * Despite the problem, IBM's SLT output for 1965 was higher than 
2- i that planned in April 1964; IBM was also able to achieve a 74% 

| increase of production in 1966 over 1965. (Knaplund, Tr. 90943-46; 
25 J E. Bloch, Tr. 91917.) In May 1966, the 2,000th System/360 was 

! shipped. (JX 38, If 27, p. 10.) 

1 

'I -371- 



r 
i 

r 
3 ; 

i 

?! 

" i 

a j 

i 

3 | 

i 

4>; 



hire and train many new employees. The size of the job was com- 
pounded by the software difficulties with OS/360 ► IBM placed a 
"top priority" on the solution of those problems and, at its peak, 
M' had over 1000 people working on. OS/360 ~ Some 5000 man-years went 
into its design j construction and documentation between 1963 and 
1966. (PX 468, p. 31? DX 13677, p. 7; see also DX 4740 : Evans, Tr. 

(Telex) 3932-34.) 

The breadth and complexity of System/360 together with 
the new, advanced applications for which it could be used required 
IBM to provide "the most extensive total programming systems support 
ever developed". (DX 13677, p. 7.) It also meant that IBM would 
have to provide customers more assistance than ever in installing, 
understanding and applying 360 and all its revolutionary new concepts. 
(Case, Tr. 73590; Evans, Tr. 101127-28; DX 1172.) 

The need to expand quickly to meet the unforeseen explo- 
s : sion in demand for 360, to hire and train new employees and to 
it support customers' in their installation and use of the new systems 
'\ placed "tremendous capital demands" on IBM. (Cary, Tr. 101525-26; 

>3 



it 

_ 1 DX 8886, p. Ill; DX 13677, p. 5; DX 13678/ P- 7-) During 1964 IBM 

:| had prepaid $160 million in debentures and promissory notes. (PX 

•! 

„ i 5771, p. 36) As a result, it did not have sufficient money on hand 

/ 2. i! 

'■i to finance the required expansion and had to raise it. In 196 6 IBM 

:| raised approximately $371 million through an eauitv offering, the 

1 first such offering since 1957. (DX 13339, p. 20; DX 13678, p. 39.; 

:J IBM Chairman T. J. Watson, Jr., explained to IBM's stockholders: 



-372- 



L 
Z 

z 

4-! 

5 

7 
3 

ii 

12 
13- 
14 
15 
IS 
17 

is! 






21 ! 



"Because of the plant construction program and the 
System/360 production build-up, 1966 required a 
worldwide investment, of approximately $1.6 billion in 
rental machines and parts, factory, laboratory and 
office equipment, and land and buildings. To help 
finance this expansion in our business, additional 
capital stock was offered to stock holders last June. 
$371 million of new capital was raised in this 
manner."* (DX 1367 8, p. 7.) 

In 1966 and 1967 IBM raised its lease prices and decreased purchase 
prices by 3 percent. (PX 4481A, p. 1.) A "major consideration" 
for the change was "to encourage purchase, and thus, increase the 
amount of cash needed to finance higher-than-anticipated demands 
for the 360". < PX 6153, p. 2; Cary, Tr. 10X525-26.) 

IBM's multi-billion dollar investment yielded fantastic 
rewards , changing the face of IBM and of the computer industry for 
all time. Chairman T. J. Watson, Jr. called 360, at the time of 
announcement, "the most important product announcement in company 
history". (PX 1900, pp. 7-8.) He could not have been more right. 
System/360 was a "phenomenal success", perhaps the greatest "in the 
history of American industry". (Cary, Tr. 101781.) As we have 
already seen and as IBM's current Chairman, Frank T. Cary, testified, 
"customers loved it", and "ordered it in quantities way beyond 



In 1965, IBM had reported that "the plant expansion program 



i and System/360 production required a record worldwide investment of ! 

22 '.» $1.1 billion in 1965 for rental machines and parts, factory and i 

office equipment, and land and buildings' 1 . (DX 13677, pp. 6-7.) [ 



25 



-373- 



I 

3 

7 
3 

3 



I 1 



anything that we had anticipated". (Tr. 101781.) 360 shipments 
exceeded by more than double the estimates made prior to announce- 
ment. (Case, Tr. 73258.) 

The effect on IBM was profound. At year-end 1963, when 
the production buildup for 360 was begun, IBM employed 138,281 
people worldwide (PX 5771, p. 3); by year-end 1969, IBM's employ- 
ment had nearly doubled — to 258,662. (DX 3364, pp. 3-4.) Over 
that same period, IBM's manufacturing floor space in the United 
States climbed from just over six million square feet to more 
than fourteen million square feet — more than double. (DX 1396 3 
DX 13964 , pp. 1-3.) At year-end 1965, before volume shipment of 
360 had begun, IBM had worldwide revenues of $3,572,824,719 (DX 
13677, p. 5); by year-end 1970, IBM's worldwide revenues had 
increased more than two times, to $7,503,959,690. (PX 5767, 
p. 3.) Just prior to the 360 announcement, IBM had approximately 
11,000 systems installed in the United States. By the time 370 
was announced, that number had tripled to approximately 35,000. 
In the interim IBM's corporate growth, revenue and profits were 
"way beyond anything that [IBM] had anticipated". (Cary, Tr. 101360, 
101781; DX 4740, Evans, Tr. (Telex) 3934-35.) 

These numbers demonstrate the extent to which IBM's 
jj success, as it stood on the threshhold of the 1970s, was the 

i 

I result of an overwhelming acceptance by users of System/3 60 and 



I 



-374- 



Z 

z 



4»| "Observers have characterized the 360 decision as perhaps 
the biggest,, in its impact on a company, ever made in 
American industry — far bigger even than Boeing ' s decision to 
go into jets, bigger than Ford's decision to build several 
million Mustangs. 



3 

e 

7 

a 

la 

u 

IZ 

14 

15 
Iff! 
17 '; 

» I 

20 1 

21 ij 

1 

22 ;! 



of IBM's ability to put the system into production and install it 
in unprecedented and unforeseen numbers. As T. V. Learson wrote 
in October 1966: 



n i 

2± 

25 



"IBM. has certainly not been the same since, and never 
will be again". (PX 1900, pp. 8-9.) 



-375- 



36. Initial Competitive Responses ■ to System/360 . The 
System/360 announcement and its subsequent success provoked a host 
of competitive responses from a variety of different sources, including 
systems suppliers, leasing companies, peripherals manufacturers and 
[software houses. As we discuss below, System/360 spurred the rapid 
growth of leasing companies, software suppliers and peripherals 
suppliers in particular, and each applied increasing pressure on IBM 
as they grew in strength. In this section, we review the more 
immediate actions taken by a number of systems suppliers. (These 
actions are discussed in more detail below, pp. 377-84.) 

We saw above how, by the time of the System/360 announcement, 
IBM's earlier computer lines had been "leapfrogged" by competition, 



L 
I 
I 

3 
9 

a 

x 

•7 | 
"" I 

-» | and how System/360 gave IBM a price/performance advantage over 

competitive machines. Indeed, as Knaplund testified, it was understood 
that the price/performance advantage of System/360 as measured by 
! IBM employees understated the true superiority of System/360 compared 
to competitive offerings.* (Tr. 90503-05.) The System/360 announce- 
ment, therefore, forced IBM's competitors to reduce prices or 
j increase performance in order to remain competitive. Weil of GE 

i 

said in June 1964: 



J I 

,3 

3 



| "The entire competitive picture in the information 

II ; | processing business at this time in 1964 is characterized 
\ 

2 :! 



£3 ;• * According to Knaplund, the methods available within IBM at the 
time for making price/performance comparisons could not adequately 

i- (evaluate several advantages of System/360: the use of disks, the 
! improved reliability, the factor of compatibility and the software 

If I support. (Tr. 90504-05, see Tr. 90506-09.) 

.i 

I -376- 



Li 

2 j 

z\ 

I 

3 I 
| 

7 I 
3 I 
9 I 

La-; 
II 

IZ 

n. 

LO- 
IS 



by the impact of the IBM System/360 . . . announcement and 
by the reaction to this announcement of our competitors. 



"The System/ 3 60 is an excellent product line with out- 
standing peripheral offerings." (PX 320, pp. 12-13.) 

The result, according to Weil, was that it was "no longer possible 

to offer equipment with a significant advantage over IBM". ( Id. , 

p. 14.) In July 1964, Learson reviewed the price reductions in the 

industry that had taken place since the System/ 360 introduction and 

wrote: 

"There, can be only one conclusion; namely, the cost/ 
performance of computers today is less than it has been 
and . . . the price structure surrounding the main body 
of our line is threatened by: (a) Present day cost[,] 
(b) New technologies, as typified by NPL[.] Perhaps 
what we are missing is that NPL was a price reduction of 
30-50%, so that competition is forced to come along with 
us." (DX 1525.) 



And they did, with price reductions, product announcements or both. 

In order to be competitive, most companies tried to price their 
la ;l products to achieve anywhere from a 5 to as high as a 40 percent 
17 ij price/performance advantage over IBM's 360 line. Despite the 
IS ;! acknowledged difficulties of comparing the performance of systems 

19 1 (see, e.g., Palevsky, Tr. 3269-71; McDonald, Tr. 4207), such a 

20 ;l pricing policy was common among competitors , who felt they needed to 

21 -j offer something better than IBM to attract customers. (See below, 

■i . 
22,'tpp. 377-84) IBM monitored these reactions in some detail, and undertook 
1 

22 • to respond, 
i 

2^ 'j a. RCA . RCA both reduced prices on its current products 

2f [and shaped its planned new announcements in reaction to 360. 

\ 

} -377- 

j 
i 

■i 



L 
I 

I 

M 

2 

T\ 
3 
9 ! 

a 

i \ 

2 i 

i 

f: 






17 ;: 

TQ 

20 



In approximately May 1964, according to internal IBM reports, RCA 
reduced the price of its 3301 between 20 and 35 percent. (PX 2956, 
p. 1; DX 1525; see also PX 4829, p. 19.) Within IBM the price 

reductions were seen as "drastic", as "the first significant com— 

« 

petitive reaction to System/360" and as making "the 3301 very 

competitive in the model 40/50 area". (PX 2956, p. 1*)* Withington 
wrote that the "primary reason for the price reduction . . . would 
seem to be a requirement for a competitive product during the interim 
until RCA announces its 'counter- 3 6' * efforts". (PX 4829, p. 19.) 

Soon after, RCA announced the Spectra 70 Series, which was 
designed to be compatible with the 360 line. (See below, pp. 551-58.) 
The preliminary design of that series had started in 1963, with 
"[mjajor design efforts. . . . under way by the latter half of '64". 
(Beard, Tr. 8459; see p. 551 below.) The strategy of compatibility 
with IBM equipment had been considered prior to the 360 announcement 
(Beard, Tr. 9113-14), and was firmly decided "within two weeks, 
three weeks at the most, after the announcement". (McCollister, Tr. 
9630.) By making its Spectra 70 compatible with IBM's System/360, 
RCA hoped to be able to persuade 360 users to move to Spectra: it 
was "aimed primarily at the IBM 360 series range of computers". 
(Beard, Tr. 8459; see pp. 552-58 below for a fuller explanation of 



21 | 

j this strategy.) 

..1 

I 

I * In July 1964, Learson interpreted the 3301 price decrease as 
\X something forced upon RCA by the "bad price/performance ratio" of the 
3301 product and its failure to sell. (DX 1525.) 



-378- 



3 

10 

II 
12 
13 

L£: 

15 

is i 

17 1 
IS 



L ; Within IBM the announcement of Spectra was noted in a 

Z l memorandum from C. E. Frizzell, President of GPD, to T. J. Watson, 
3 I Jr. Frizzell wrote that the series offered better price/performance 
4* |[ than IBM in CPU -memory speed, magnetic tapes and high speed printing 
y- |t but assured Watson that he was ,r moving rapidly to meet this challenge 
3 : and expect to respond effectively in the very near future" . (DX 
7 If 960.) 

b. GE. Weil testified that at the time of the 360 
announcement, GE assessed 360 as a "very strong and very competent" 
competitor to its current system, the GE 400. (Tr. 7060.) As 
noted earlier, his own assessment in 1964 was that it was "no longer 
possible to offer equipment with a significant advantage over IBM". 
(PX 320, p. 14.) Then he reported at the same time (June 1964) that 
GE was planning to announce a new series of magnetic tape units 
"which will permit adjustment of our 400 line system prices to 
increase our competitiveness". ( Id. , p. 16.) IBM sources reported 
that GE did reduce prices on the 400 in reaction to System/360. 
Lear son wrote in July 1964: 



• "GE has not officially reduced prices, but they 

IS j|. are selling their 400 line at 18% off. They have also 
reduced their extra shift to a 10% charge. 



2° 

:i "Further, GE is selling their 635, a competitor to 

21 •[ . the 7094, at no extra shift charge." (DX 1525.) 

22 !| A September 1964 Competitive News Release from the Data 



»i 



23 | Processing Division's Commercial Analysis ' Department confirmed 
24- } price reductions of 8%-15% and went on to say, "The price 

25. ' 

.i 

•I -379- 



I i 
z 

3 

a i 

6 

7 

3 

9 

ia 
ii 

12 

13 
14 
15 

is ; 

17 i 

! 

18 ■ 

i 

19 ; 

20 ; 

21 I 
22 

24 
25 



reduction gives the GE 400 a price/performance advantage over com- 
parable System/360 configurations." (PX 2966, p. 3*) However, 
Knaplund felt, that the price reductions were necessary for GE to 

remain competitive after System/360. ( Id. , p. Rl; see below, pp. 
490-93.) A subsequent price/performance evaluation made within IBM 

concluded: "While the recent price reductions have improved GE's 
position, the System/360 Model 30 retains its price/performance 
superiority." (DX 13445.) 

General Electric announced its 600 series in the summer of 
1964. Although planned long before the 360 announcement to displace 
IBM's 7090 and 7094 computer systems (see below, pp. 493-505), GE 
called the 600s a "family . . . for business, scientific and real- 
time use". (DX 491, p. 1.) Weil had compared the 600 series against 
the 360 line in a June 23, 1964 internal GE presentation and con- 
cluded that the 600 is "either just a little more favorable or just 
a little less favorable than comparable members of the 360 series. 
We are, however, able to deliver our equipment a year earlier than 
IBM". (PX 320, p. 16; see below, pp. 493-505.) 
!j GE saw itself as being able to capitalize on one of the 

j risks IBM had taken with the 360 — the risk involved in making the 
:l older lines obsolete. Weil testified that the computer group at GE 
jl was "initially at least overjoyed with what had occurred because it 
meant right at the time we were introducing a system designed to 
displace 7090s and 7094s, IBM had itself abandoned 'che 7094 and 7090 
computer series and brought out an entirely different computer series, 

-380- 



L 
2 
3 

4. 



and it was our belief at that time that it would be easier, if you 
were a user, to convert from the 7090/7094 to the 600 series than it 
would be to convert to IBM's new 360 series. We regarded that as a 
fortuitous occurrence and potentially to our advantage." (Tr. 7060-61.) 
The user of the 7094 was "forced ... to either go to a 360 or to 
some other competitive system, and we were sitting there with a system 
designed to make that conversion as easy as possible." (Tr. 7062.) 

c. CDC . According to Weil, CDC also reduced prices in 
response to System/360. (PX 320, p. 16.) At IBM, Learson analyzed 
CDC ' s behavior as follows : 



7 
3 
9 

ia 

Ll 
12 
13 
14- ; And Withington wrote 



"CDC followed [360' s pricing] with a price reduction 
of their 3600, which was no longer competitive with the 
360-Model 70. In dropping the price of the 3600, they 
had to keep their deck of cards in order and so moved the 
3200 and 3400 downward. Reductions of 20-40% were made." 
(DX 1525.) 



"Control Data's main reliance is on price; apparently 
its intention is to provide a lower cost answer to every 
IS '! System 360 model. After the System 360 announcement, the 
;i price of every existing Control Data computer was reduced, 
U ;| and the prices of the later models are still lower. . . . 

This should unquestionably help Control Data's position 
^g J because . . . the market is becoming increasingly price- 
conscious." (PX 48 29, p. 21.) 



:! 

:| 

19 I 



Several months later, CDC announced new members of its 



•] current product lines — the 6000 and 3000 series. The formal 

21 i 

j announcement of the 6400 (a "scaled down" 6600) and the (never 

22 ; S 

j delivered) 6800, to go with the existing 6600, was made in mid- 
23 -i 

(December 1964. (Norris, Tr. 5626, 5965-67; DX 319, p. 1.) The 3300 

~" jand 3500 were announced in 1965. (PX 355, p. 35.) 

I 
I. 
i 

! -381- 



z 

3 

4. 

3 ! 
i 

7 I 

3 

a 



3 



d. S perry Rand. Sperry Rand came out with new products 

in short order after the announcement of System/360. Two weeks 

after the 360 announcement, Univac management met to consider the 

Univac Product Line Strategy. They decided to enhance and expand 

the 1050 program to provide a compatible line of systems from the 

1004 through the 1050 Mod V. (DX 14, p. 1.) Learson reported in 

July that Sperry was "announcing new models of 1050 and 1004 where 

the price/performance ratio is not following the historical trend in 

the original announcement, so they are, in effect, using this as a 

method of price reduction". (DX 1525.) Univac management also 

decided to extend the 1107 program to the 1108 and 1109, which were 

to be program compatible upwards with the 110 7, for large scale 

:j users. (DX 14, p. 1.) In mid-1964 Sperry Rand announced its 1108 

.i 

;| at a price which Withington described as "impressive when compared to 



■that of the System 360". (PX 4829, p. 20.) Withington wrote that, 

"^ I 

(in terms of price/performance, "IBM's initial offerings in the 360 

•6 :! 

:| line were inferior to it". (PX 48 30, p. 22; see below, no. 477-30.) 

: .7 ; 

:} (We shall see later IBM's response to this rather quick "leapfrogging".) 

• S 

: By 1965, Univac ' s Product Line Task Force was contemDlating 

:si| 

:j the introduction of an entirely new product line in reaction to 

System/360. It faced a dilemma in that two of the three models 

II ] - 

j under development were likely to benefit from new technological 

22 "> 

t developments if their development could be delayed, but waiting 

23 • 

I would have meant that a full family could not be announced at one 
i 
2± 1 

jtime. (DX 16, p. 2.) Univac finally compromised and announced the 



-382- 



3 
9 

ia 

11 

12 
13 

L4 



Li 9200 and 9300 (rather than an entire family). (McDonald, Tr. 3821; • 
Z DX 70, p. 9.) These systems "aimed at compatibility" with 360 
3 ; (Eckert, Tr. 908) but achieved it only in part. (See below, pp. 480-86.p 
4*j- e. Burroughs . Burroughs also responded with a new 

product introduction. In August 1964 Burroughs announced the B5500 
(PX 2082, p. 95) , "a more powerful successor to the earlier B 5000", 

"i| 

j ;} and what was to become the first member of the 500 System family. 

(PX 4829, p. 22.) Withington described the B 5000 family as "incor- 
porat[ing] very advanced design features, facilitating the use of 
compilers and executive programs", but he concluded that "Burroughs 
apparently has not attempted to answer the System 360 across the 
board". ( Id. ) By 1966 Burroughs had turned the 500 family into "a 
major new product line" (PX 4832, p. 21), adding the B 6500, 2500 
and 3500 to the 5500 and the very large (and never delivered) 8500. 

( Id. ; DX 10262, p. 8; see below, pp. 644-50.) 

f. Honevwell. After the 360 announcement, Honeywell took 

IS, ' 

■ its successful 200 system and turned it into a compatible "family of 

17 || 

[computer systems": the 120, the 1200, the 2200, the 4200 and the 

IS : 

8200. (DX 13849, n. 27; see below, on. 619-29.) Honevwell also 

19 | 

j abandoned its attempts to develop a mass storage system after the 

2Q ] 

; 2311 introduction and began buying disks OEM. 

21 | 
j g. SDS . SDS announced successive new products beginning 

22 :» 
■j in 1964 with what it termed "the first computer to use monolithic 

n \ 

[integrated circuits, the SDS 92" (DX 44, p. 5), and eventually, the 

2d ] 

j Sigma series, which was announced beginning in 196 6. (Palevsky, Tr. 

25 j " 

-383- 



3223-24; see below, pp. 703-04.) A press release at announcement stated 
that "Sigma . . . represents the first family of computers with an 
entirely new design since the IBM 360 announcement" (DX 52, p. 1), 
L.l-and, as IBM had done with 360, SDS stressed the new line's universal 

i 

applicability. (See below, pp. 704-05.) 



j ; 
i ; 
T 
3 



2, i 



<L 



'A 

•t 

.0 '! 



2 "» 



... t 



M 



-384- 



L 
2 
Z 

3 



37. IBM's Responses (1964-66). With competitors responding 
rapidly to the initial System/360 announcements, IBM was soon faced 
with the need to respond in turn or lose the competitive advantage it 
had obtained by the introduction of System/360. It chose to respond. 

IBM did so by introducing new products , improving existing 
products and lowering prices. This section discusses IBM's initial 



7 ;j responses,* particularly IBM's reduction of extra shift charges, 
improvement of memory speeds , announcement of improved tapes and 
disks , introduction of the Model 20 , and development of the Models 
44, 67 and 90. 

a. Reduction of Extra Shift Usage Charges . At the time of 
the System/360 announcement, IBM was charging its rental customers a 
flat rate for 176 hours of computer use per month — the Monthly Avail- 
ability Charge, or MAC. For use beyond that number of hours, an 
, additional use charge was billed at a rate of 40% of the per-hour MAC 
rates. (DX 14295, p. 44.) 

One of the ways that competitors responded to 360 was by 

j reducing or eliminating charges to customers for using machines on 

i 

j extra shifts. An IBM Wins and Loss Report for June 1964 cited "erosion 

is] 

; [ of extra shift" as one of the most significant aspects of competitive 
2Q . 

:i announcements since System/360. (DX 13824, p. 2.) On July 29, 1964, 

21 'j . 

•i Learson wrote that GE had reduced its extra shift on the 400 line to 

22 > 

— 4 



3 

9 

10 

il 
12 

13 • 
14 
15 
IS 

17 ; 
is 



i * IBM's planning for its future products is discussed later. (See 
Z - \ below pp. 378-922. ) 



25 ■ 



t 

i -335- 



Li 10% and was offering its newly introduced 635, "a competitor of the 

*! 7094" , with no extra shift charge at all. (DX 1525.) 

i 

I i IBM reduced its additional use charge from 40% to 30% on 

Vil- August 11, 1964, effective retroactively to July 1. (DX 13823.) 

It was not enough, and IBM received pressure for additional reductions. 

5 ■! On August 13, 1964, Evans and others in IBM were notified by DSD's 



' 



3 : 
9 

a 

1 ! 



7 i| Advanced Systems Group that: 

"We are currently facing severe competition in the 
. medium and large scale scientific areas from such machines as 
the GE 625, GE 635, PDP-6 , etc. A goodly part of this problem 
is due to our additional use charges. GE, particularly, is 
offering their 600 series on a 24- hour basis. Even in cases 
where we are price competitive on a single shift basis, we 
rapidly become non-competitive when additional use is involved. 
The 30% extra shift charge is good but not nearly enough." 
(DX 13640, p. 1.) 
il 

In addition, IBM was losing orders to the Honeywell 200, 

particularly at service bureaus. In October DPD "fought" for a 
reduction in extra use charges to 10 percent, this being, as Cary 
wrote to T. J. Watson, Jr., in the beginning of December, one of "the 
instances where we have 'screamed' for action". (PX 1265, pp. 2, 4.) 

On October 14, 1964, IBM announced a further reduction in its 
extra shift charge for System/360 to 10%. (DX 14134.) 
| b. Memory Improvements . Within two months after 360 was 

i announced it became clear that the memory speed of certain IBM systems 
11 I . 

\ had been surpassed by newly announced competitive machines.* A June 
*1 i 



.X : 



- it 
.0 « 



.7 
.S 



.3 I 



"i * 



Even before the announcement of System/360, IBM had been seeking 
li, | to include faster memories in certain IBM computers. Evans testified 

J that, for the Model 60, a 3/4-microsecond memory, which he called 
lz \ "startling in test", had been planned. However: 



-386- 



L 

2 

3 

4- 
5 i 
5 
7 

a 

ia 

u 

12 
13 
14- 
15 

T - 6 , 

17 I 
IS • 



1964 Wins & Loss Report cited "the fast memory speeds of [competitors'] 
new systems" as one of the "three most significant aspects of competi- 
tive announcements".* In particular, the memory speeds of the 
Honeywell H-2200, the NCR 315 RMC, the Univac 1103, the GE 635 and the 
CDC 3800 were mentioned. (DX 13824, p. 2.) 

Haddad, then Director of Technology and Engineering, 
addressed this problem further in a July 28, 1964 letter to Vice 
President and Group Executive Gibson: 

"I am becoming increasingly concerned over the possibility 
that some of the 360 machines will be technically obsolete 
before they are delivered. With the recent round of price- 
cutting by some of our competitors, it is even more important 
that our machines remain technically superior. 

"There is obviously a strong trend toward the use of 
faster memories across the board. This is exemplified by 
the Univac 1108 . . . the NCR 315 . . . the CDC 3800 . . . the 
H2200 . . . and the RCA 3301. . . . All of these examples 
appear to give the competitor a memory speed advantage at an 
equivalent 360 machine level." (DX 13825.) 

The need to improve memory speed, and with it processor 

price/performance, was particularly acute for the larger models of 



"As we proceeded down the 360 development program, and so 

| to make certain that we could deliver what we were committing, 

IS ;j we decided in 1963 or early 1964 to use available memories that 

I were technologies that were proven and memories that had been in 

2C j production. And so instead of one Model 60 with three-quarter 

'l microsecond memory, we made 2 models at that performance range. - 

21 \ A Model 60 with a two microsecond main memory, and a Model 62 

j with a one microsecond main memory and that's what we announced." 

22 ;l (Tr. 101111.) 

23 j Similarly, the Model 70 was announced with * one-microsecond main 
j memory because the 3/4-microsecond main memory was not yet fully 

24 1 tested. (Evans, Tr. 101112.) 
.i 

25 i * The other two were the "magnitude of price cuts" and the "erosion 
j of extra shift". (DX 13824, p. 2.) 

1 -387- 



r i 
3 It 

: 

1 1 
z\ 

.4 

5 

.6 ii 

.7 

a 

s 



the 360 line, the 60, 62 and 70. An IBM "Wins and Loss Report" for 
August 1964 reported that "there have been no credited orders for 
Models 60, 62 and 70 since June and only a few in the uncredited 
category"^ (PX.3630, p. 2.) f 

Within- IBM it was believed that the Models 60, 62 and 70 
compared particularly poorly with CDC's new entries. On October 19, 
1964, Ralph A. Pfeiffer, Jr., then Vice President and Federal Regional 
Manager for DPD, wrote to Cary comparing IBM's models to the CDC 
6800. (The CDC 6800 had not yet been formally announced and was 
never in fact delivered) . He stated his belief that "our model 70, 
with a little less than half the performance of the 6600, rents for 
approximately the same amount" and recommended "that DPD request a 
100% performance improvement in the Model 70 with no increase in 
rental price and not more than a 20% increase in purchase price" . 
(PX 1214.) On December 1, Cary recommended that the price of the 
one-microsecond memory on both the Model 70 and the Model 91 be 
reduced "in order to make our bids . . . more competitive from a 
price/performance standpoint". (PX 125 6) DX 14504).) Those price 

j reductions were announced on December 23. (JX 38, pp. 329-30; JX 10, 

I App. A, «(<[ 3,5, dd. 2-3.) 

•j That day, DSD President Kennard wrote to A. K. Watson: 



t 
■i 

22 \ 

i 

aj 

2A 1 



"As you know, we have undertaken a number of actions to improve 
the product line and to provide specific responses to certain 
technical requirements. An example of this is time sharing. We 
have redesigned the 2362 1 microsecond memory and released new 
models reflecting this redesign. The net effect is a lowering 
of the rental and purchase price, and lower systems prices on 
S/360 Models 62, 70 and 91. 



— ' '! "We have also determined, through analysis of the requirements 
for peripheral I/O devices, that we could achieve substantial 



-388- 



L 
2 

3 

4*i 

5 

a 

7 

a 

9 

10 

II 

12 

12 

14" 
15 
is 



operating efficiencies and enhance our price performance by- 
developing a new multiplexor channel. We have done this and 
combined the multiplexor function with the already announced 
selector channel function. The net result is reflected in a 
potential systems price reduction of from $5,000 to $10,000 per 
month. When the new higher speed multiplexor selector channel 
(2870) is integrated into our plan, it is apparent that the 2860 
price for its function had to be. re-evaluated. We have completed 
this evaluation, and this has resulted in a lowering of the 
purchase and rental price. This price adjustment has been 
released to DPD along with the 2870 multiplexor channel." 
(DX 13827.) 



However, these price reductions were not enough. In 
December 1964 Kennard wrote to Watson and A. L. Williams that the 
performance of the 6400, as indicated by CDC, would place it between 
the Models 62 and 70 while " [f]ield reports indicate a price somewhat 
above our Model 50". He reported that steps were being taken to 
improve the competitiveness of those machines, including an increase 
in the memory speed from one microsecond to three-quarters of a 
microsecond for the Models 62 and 70. This and other improvement 
programs for the Model 70 and Models 60/62 were targeted for release 
in January 1965. (DX 14322.) 
17 : The competitive advantage of CDC 6400 and 6600 over IBM 

IS ■! models continued to be a concern. On March 10, 196 5, C. B. Rogers, | 

1$ ' Jr., then Director of Product Programs for DPD, wrote to Lear son: j 
; ! 

2Q j "The CDC 6600 overpowers our 70 . . . for approximately the j 

'} same rental. . . . The new entry of the CDC 6400 . . . clearly 1 

21 ! out-performs our Model 62 by a factor of 2 at a substantially j 
| lower price for both purchase and rental. ... It is accurate 

22 'j to say we are in trouble." (PX 1389*, pp. 1-2.) j 
: t j 

22 .! By April 1965 IBM was ready to announce a faster memory: I 

2i | the 750 nanosecond (3/4 microsecond) memory. On April 22, IBM announced! 

I I 

~e I the Model 65 and Model 75, each having a memory speed of three- ; 

I j 

1 -389- 



L 
2 
3 

5 
5 



quarters of a microsecond. The faster performing Model 6 5 superseded 
the Models 60 and 62, and the faster performing Model 75 superseded 
the Model 70.* (JX 38, p^ 393.) 

c. Tape Drive Improvements. Soon after its announcement 
of System/360, IBM also recognized the need to improve its peripherals 
in order to maintain the superiority it had achieved in the 1950s. On 



7 I August 21, 1964, the System/360 Compatability Committee reported 



that because of the nature of 360, peripheral manufacturers could be 

expected to market compatible replacements for IBM's peripherals: 

"(1) I/O manufacturers, whether independent or divisions 
of computer manufacturers, are in a position to market devices 
of comparable IBM capacities at approximately 20% less price. 

"(2) It appears that I/O manufacturers will attempt to 
sell tape drives and terminals to System/360 customers. 



a 

10 

12 

13 . 

ij "(3) There will probably be concerted activity from 

T # ' competitors in marketing I/O devices on System/360 in the 

~ i Federal Government." (PX 3908-A, p. 4.) 



They stated that: 



io if 



"The heretofore heavy emphasis on processor planning as 

imf \ the criterion for improved price/performance should be 

u i re-oriented towards I/O developments . The across-the- 

I board improvements in price/performance which will be 

: ^ ij required in the 1967-68 time period will probably be 

I brought about more by improved I/O capability than by CPU 

*^ ;} and memory improvements. As part of the regular develop- 

| ment effort, such activity will be necessary in any event 

-G • to keep System/360 a viable product line . . . . " ( Id. , 

I p. 22.) 

21 'I . 

•i 



?*? 



•J * IBM did not limit its memory improvements to its larger models. 
23 i on January 4, 1965, IBM announced that the memory for the Model 30 had 

j been improved, from two microseconds to 1-1/2 microseconds. (See PX 
24. i 1288, p. 2; PX 1637, p. 2; DX 14135.) 



-390- 



Technological improvement was additionally important, the 
Committee reported, because 

"competitors will attempt to market I/O devices, with particular 
emphasis on tape drives, directly to 360 users." ( Id. , p. 24.) 

Tape drives, in particular r were an area that needed improve- 
ment. A presentation to the DP group staff in November 1964 by a 
group headed by C. J. Bashe entitled "Group Staff Review of IBM's 
Technological Position in the Marketplace" summarized IBM's position 
relative to its competitors from the viewpoint of research and 
development. That presentation reflected IBM's unparalleled commitment 
to R&D and showed that IBM, in comparison to CDC, Burroughs, NCR and 
Sperry Rand, had consistently devoted a larger portion of its revenues 
; to research and development. (PX 6671, p. 5.) Nevertheless, it 
showed areas in which IBM was not ahead. The report concluded: 



L; 

z 

z 

4- 

T 
3 

10 

II 
12 
13 

*•• if "We are ahead of competition in some but not all of the 

,_ ! technology areas critically important to system performance. 

^ \ We do not have an unassailable position of leadership in any 

! function." ( Id. , p. 27.) 

! It recommended attention to "box-by-box superiority" and concluded 

•' -i 

j that half-inch compatible tape drives was an area in which IBM was 

IS : ! 

:| "inferior". (Id., Dp. 15, 26-27.) 

is ;j — * 

i A General Managers ' meeting was scheduled by Knaplund for 

ZQ ! 

i December 4, 1964, at which technical managers were expected to report 

21 :| . 

j on action plans to solve the problems in the areas in which "IBM must 

22 1 - • 

.1 take immediate action to attain technical superiority". One of the 

23 : 

■*• j 



[ topics was "[a] superior performance 1/2" tape drive to be announced 
j in 1965". (PX 1251 (DX 14503), p. 1) This was considered necessary 

j because: 

i 

'i 
» 

;! -391- 



3 

3 

a 

3 



"We're outclassed in half -inch tape and apparently can't sell 
one-inch tape equipment. We need a tape drive that is superior 
in performance and acceptable." ( Id. , p. 4.) 

A week later, on December 11, a Peripheral Task Force 
reported. This group considered the use of small systems to control 
peripherals, as in tape-to-printer or card-to-tape applications. 
Such applications were common, applications for the 1400 series com- 
puters and it was expected that the 360 Model 20 and Model 30 would 
also be used for such purposes. But the Task Force believed that IBM 
had a problem and could "expect to lose approximately 500 systems {in 
that application area] by the end of 1965 with the presently announced 
product line", with losses expected to continue thereafter. To 
minimize such losses, the Task Force recommended the announcement of 
the 2400 series tape drives on the Model 20 "immediately" and stated:. 



^ ;■ "Low Cost Tape should be announced on the Model 20 primarily to 
satisfy the longer term problem (after 1965)." 



Further : 



j 



,S 



"Low Cost Tape on the Model 30 is required to provide more 
competitively priced configurations particularly to those 
customers requiring 1401 compatibility." (PX 1271, p. 3, 
see pp. 6, 8, 12, 14. ) 



„ I IBM's fears about its lack of technological superiority in 

•S \ 

\ tapes were made even more immediate by additional actions of its 
•G ! 

i competitors. On December 11, 1964, the same day the Peripheral Task 
•i Force issued its report, C. E. Frizzell, President of GPD, reported to 
T. J. Watson, Jr., on the recent RCA Spectra 70 announcement. He 



Li ; 



?«; 



listed among the "significant advantages" of the Spectra 70: 



-392- 



L 
2 

z 

4- 
? 
a 
7 
3 

ia 
ii 

12 

13 



"One- third higher speed magnetic tape drives at equivalent 
rentals compared to IBM. . . . Availability of magnetic tapes on 
the Model 15 gives them a magnetic tape system in a price range 
where we have no current entry." (FX 1272 (DX 960), p. 2.) 



Honeywell, CDC and GE tape drives were also a problem. On 
December 22, T. V. Lear son wrote to T. J. Watson, Jr., concerning 
"out-performed, out-priced market areas". He listed "low cost tape 
systems" as one of these and stated: "This is largely the Honeywell 
200 story". . He called for "[t]apes on the 360/20 [to be] immediately 
announced". (PX 1288, pp. 1-2; see DX 13955, p. 4.) 

IBM improved its tape drives in two steps. The first step 
was the announcement of the 2415 tape drive and control unit on 
April 5, 1965. (JX 38, p. 377.) The 2415, a lower cost unit for the 
Models 20 and 30, solved the tape drive needs of users of those 
models. The second and more important step was the announcement on 

-** |j August 9, 1965, of the 2401 Models 4, 5 and 6 tape drives and control 
units. These 2401s incorporated several advantages in tape technology 

IS |j including: 1600 bit per inch density, phase encoding recording and 

17 |[ twice the data transfer rate of IBM's earlier models. ( Id. , p. 

IS ■] 484.) 

19 :| For the time being, IBM appeared to have solved its problems 

2G in tapes with the new 2401s and the 2415. (PX 4256; DX 13950, p. 2.) 

J 
21 ;( Soon, however, competition, particularly from PCMs, would push IBM to 

i 
22] improve its tape drives even more. (See below, pp. 886-90.) 



« «• 



23 s{ d. Disk Drive Improvements. As we have seen, I3M made the i 

" * i 

! i ' 

24. ; j disk drive an integral part of its System/360. (See above, pp. 323-28. )j ; 

25 J The 2311 disk drive, announced as Dart of Svstem/360 (JX 38, p. 86), ;j 

i * i 

i ■ 



-393- 



was "the first very reliable disk drive". (Beard, Tr. 9048.) Com- 
petitors initially were unable to offer a similar product. (Case, 

l\ Tr. 72744; see also Withington, Tr. 56240-41.) 

M' For several years prior to introducing the 2311, IBM had 

: marketed the 2302, which was a drum-like file with very high capacity. 

5 ji The 2302 was larger than the 2311, but with not as much versatility as 
the 2311. Soon after System/360 was announced IBM found that 



"it was beginning to be apparent that customers had a far 
greater need for data stored in disk drives than we had 
anticipated a year or two earlier when System/360 was under 
development and when the 2311 disk drive was first intro- 
duced." (Case, Tr. 72742-43.) 



.0 !« 



Thus, IBM needed a disk drive larger than the 2311 to replace the 2302 

^ : and supplement the 2311. (PX 3226A, p. 4.) 

Against that background IBM introduced its 2314 for two 

reasons: first, since the 2314 would be larger than the 2311 it 

would "provide a better relation to competition than the 2302 files". 

(Id.) Second, because the improved price/performance of the 2314 

| would improve the overall system performance of 360 systems on which 
3 \ 

"| it was used, "the 2314 was announced ... to sell more 360 systems". 

:| (Id. , p. 5.) 

.3 i 

f 

■i IBM announced the 2314 on April 22, 1965. (JX 38, p. 439.) 

i The 2314 "[vjery definitely" represented an advance over prior disk 
11 •! 

! j drives. (McCollister, Tr. 9597.) Compared to the 2311, the 2314 
| provided an increased capacity of four times per spindle (Case, Tr. 

i 

j 72742) , an improvement of two times in data rate ( id. ) , and the 

2- "! 

■I ability to operate on-line. (PX 1967 (Tr. 35690).) Beard, who then 
lz \ 

1 worked as Chief Engineer of RCA's computer division, testified that 

j -394- 



while "the 2311 demonstrated the reliability" of random access devices, 

"[t]he 2314 not only offered the reliability but also a practical 

cost for the random access user". (Beard, Tr. 9049.) 

The superiority of the 2314 provided substantial benefits 

to IBM. The 2314 "turned out to be very surprising in the rate that 

customers found use for it" • IBM "totally underestimated the demand 

for such devices" and was "hard pressed to deliver the devices as fast 

as customers were demanding them". (Case, Tr. 72743.) It also had 

the desired effect on systems sales: 

"The availability of the 2314 has been the catalyst to make 
many systems sales for previously undeveloped application use 
of computers." (PX 1967 (Tr. 35690).) 



L 
2 
3 

a 
7 
3 
9 

La 

LI 
IZ 
13 

L4- 
15 

L7 !| 

•j "During that period the entire industry and the users began 

^3 !| to appreciate the importance that disk drives were going to play 
;{ in the great majority of general purpose computer systems. . . . 
1$ ; I believe only IBM among the major competitors at the time 

; offered an alternative between magnetic card devices and disk 
20 j drives, with developments proceeding along both lines. . . . 

j When it became apparent that the class of magnetic card devices 
Zi | was not going to be successful in the marketplace, for reasons 
of reliability, and that the disk drive was a critical product, 
many of IBM's competitors were left* for awhile without a satis- 
factory option." (Tr. 56240-41.) 



And: 

"The 2314 is an example of where the product developed a 
market beyond our initial forecast expectations. Every company 
should have a door-opener that beats competition — the 2314 is 
such a product and will continue to be only if our pricing 
policy can stand the challenge of competition." ( Id. (Tr. 35692).) 

Withington echoed the advantage which the 2314 gave IBM over 

its competitors : . 



22 : 

23 : 



25 



e. Introduction of the Model 20. IBM's success with its 
650 and 1401 had shown that small, low cost computers were important 
because they helped grow the market by permitting users who otherwise 

-395- 



r : 



L 



might have been unable to afford them to obtain computer systems. ( See 
above, pp. 39-44, 141-47.) In the face of that experience, the SPREAD 
Committee had recommended that IBM develop a "very small" processor, 



Mf even though such a processor' might not be fully compatible with the 



r 

2 

7 
3 
? 

a 
i 
z 

A: 

.6 , 

,7[ 
.3 ■ 



rest of the 360 line. (DX 1404A, pp. 35-36, 69 (App. A to JX 38).) 

The development of such a small processor was assigned to the World 

Trade Corporation's German laboratory in Stuttgart. (Hughes. Tr. 

71942-43; Knaplund, Tr» 90478.) In early 1964 that small processor 

was judged "not to be as far advanced in development as the Models 30 

through 70", and it was therefore not announced with the rest of the 

360 line in April 1964. (Knaplund, Tr. 90489.) 

The need for a low-cost computer was evident within IBM. A 

document of April 15, 1964, entitled "Forecast Assumptions for the 

1430N Data Processing System" (the 360/20) , stated: 

"The 1430N Data Processing System will offer the 
advantages of stored programming to customers and 
prospects for whom mechanization of data processing 
has heretofore been either impracticable or confined to 
conventional punched card equipment. 



"The 1430N, which will have a subset of the NPL 
instruction set, will be the smallest member of the 
;j System/360, and will benefit strongly from the impact 
'£ | of the recent announcement of the NPL line. 

V3 ! "This system will bring the world of the System/360 down to 

■} the price range the small user can afford. 

!T ; | 

j "For the first time a new technological breakthrough, 

12. \ like the one realized with SLT for the System/360, will 

,j be made available at lower cost to the small customer at the 

23 ; same time as to the larger user. 



2- i "The 14 3 ON system offers growth within the system and 

| upward growth into the System/360, Model 30." (DX 13829, p. 1.) 



-396- 



L 
2 

3 

4*i 
5 
a 
7 

S 

a 

lq : 
11 

12. 

13 



Even though, as this suggests, the 360/20 was in large part 

expected to be acquired by new users , it was anticipated that a 

variety of customers would find it attractive, including: 

" (1) Small companies characterized by one accounting 
machine installations [sic] . 

"(2) New customers in this size range* 

" (3) Larger unit record customers who have not yet moved 
to a system. In some cases, these customers will use multiple 
1430N systems. Others will use a 1430 along with some unit 
record equipment. 

" (4) Users of large data processing systems who still 
have a considerable amount of unit record equipment installed. 
In these cases, -the 1430N would replace some or most of the 
unit record equipment supporting the larger systems . 

"(5) The Communications Market. This market will 
be characterized by customers having a number of branch 
locations requiring frequent and/or prolonged contact with 
the central data processing center or among each other." 
(Id. , p. 2.) 



After the announcement of System/360, the need for the 
Model 20 increased. On July 20, 1964, Opel wrote to Learson concerning 
- 3 .'banking product deficiencies and stated that "[w]e need to have a 
17 ;| more competitive response to the [Univac] 1004 and other competitive 
IS :| small card processing systems". (DX 14477.) Writing after the 
IS {announcement of the Model 20, Withington observed that IBM had to 

20 | announce "such a computer to protect its position" from "the Univac 

21 11004 and 1005, the Honeywell 120 and the GE 115". (PX 4830, p. 20.) 
•i 

22 i IBM announced the Model 20 on November 13, 1964. (JX 38, p. 

22 ;296.) Because of the need to keep its cost down, the processor did 

24. 1 not share all the features of the 360 line. First, the Model 20 

25 j contained only a subset of the 360 instruction set and, hence the 
i 

! 
i 

=| -397- 

'i 

i 
i 

i 



L 

r 
j 

T\ 
I 



Model 20 was not compatible with the rest of System 360 to the 
extent that the other members of the System 360 line were compatible 
with each other ► (Case, Tr. 73575; JX 38 , p. 297.) Second, the 
360/20 did not use the System/360 standard interface for attaching 
peripherals and instead used "native attachment". (Hughes, Tr. 
71992; Case, Tr. 74085.) This was done so that IBM could offer the 
360/20 at a more competitive price : 



"In order to achieve [the lowest possible price] , you have to 
have the lowest possible cost to the manufacturer, and in order 
to achieve that it is sometimes necessary and was in the Model 
20 necessary to design a special unique means of attaching disks 
to the Model 20 , because using the standard interface for that 
purpose on the Model 20 would have been more expensive and would 
have therefore unnecessarily increased the price of the Model 
20." (Case, Tr. 74085.) 



£ 



j The Model 20 announcement stated that it was "a System/360 

for card processing ... a stored program approach for smaller 
business needs". (JX 38, p. 296.) As noted above, however, IBM's 
forecasts for the Model 20 anticipated many potential users other 
than small users. The 360/20 was in fact used in a variety of ways 
7 '>! by a variety of users. For example, an IBM Competitive Daily Report 

■2 i| stated that "[tjhere are about 600 Model 20' s installed with communica- 

a ;| 

•* j tions equipment and 700 installed in large customer accounts". 

•r "' I 

^ .} (PX 3773, p. 2.) And Wright testified that a sample configuration of | 

1 ! 

" • a multiprocessor Model 67 system contemplated the use of 360/20s in \ 

j j 

£► is connection with the Model 67 in various ways including as concentrators i 

i- ] I 

^ j for terminals. (Wright, Tr. 13348-49.) Similarly, DX 4851, a memo- 

j i 

l± "s randum on the GUIDE Project on Remote Batch Computing of February ! 

25 I ■ ; 

i 

■! -398- 



L 
2 

111 
4* 

5" 

7 
3 

10 

II 
12 
13 
14 
15 
Iff 
17 
IS 



1966, contemplated a large 360 as "the central facility" of a system 
in which "[t]he remote terminals may be small typewriter keyboards; 
but, more likely, will be Model 20 360' s, Model 30' s, or even larger 
machines with their own operating systems". (DX 4851, p. 5.) "The 
bulk, of the terminals planned for use would be small computers [ , ] 
mostly 360' s Model 20 or 30." ( Id. , p. 6.) 

As it turned out, the 360/20 was more than merely a good 
competitive response; like the 1401 in its day, it became the largest 
selling of the 360 systems with more than 7400 installed in the United 
States by 1970. (DX 2609B, p. 182.) 

In December 1965, Withington summarized the effect of IBM's 

competitive responses: 

"Soon after the System 360 line of computers was 
announced, it became apparent that despite the basic 
soundness in the line there were a few deficiencies and weak 
points. IBM, apparently desiring to establish a product position 
now that will remain sound for a number of years, has moved very 
vigorously to remedy the deficiencies. It has announced new 
products to add to the line, improved the price-performance of 
the initially announced products, and adjusted marketing policy 
in certain respects. 



it 

;| "The 360/20 extends the line downward in price, while still 

IS t retaining most of the features of a full-scale computer system. 

;j Considering the appearance of the Univac 1004 and 1005, the 

ZQ i Honeywell 120 and the GE 115, one had to expect IBM to announce 

:\ such a computer to protect its position in a market area repre— 

ZL ;j senting important dollar volume. It should be effective protec- 

.j tion; the 360/20 offers very competitive price-performance 
characteristics. ... 






x 

21* \ "The 360/65 appeared when it was clear that the initial 

;j 360' s at the 'top of the line' could be bested by the competition 

25 I The 360/65 cannot, at present; it offers price-performance as 
good as anything on the market. . . . 

't 

•I -399- 



"When pressed by competition, IBM has also made significant 
improvements in the previously announced products — even before 

I ; delivery of the first models. The 360/30 initially showed a 
price-performance characteristic inferior to those of some of 

Uj its competitors, so IBtf increased its speed sharply by substi- 
tuting a faster memory at no increase in price. The initial 
terminals and control devices for remote input-output were too 
expensive, so IBM has supplemented the initial offering with a 
number of lower-cost devices ► Perhaps most important overall, 
IBM increased the packing density of all its magnetic tape units 

j\ from 800 characters per inch to 1600, at a small increase in 

•j price, by using a new recording technique. This factor is 

a : important to the overall productivity of most computer installa- 
tions, so the entire 360 line benefited considerably^ The 

i I competitors will be able to match this improvement, but for the 
time being IBM's position is improved." (PX 4830, pp. 20-21.) 

1 ' 



i 

i 
S ; 

7 ; 
si 



i 

— I 



La ■■'■ 
i 

u \ 

I 



-400- 






7 

a 

a 

10 

II- 
12 
13 

15' 
lo 



f. The Model 90 Program. As we have already discussed, 
during the 1950s IBM undertook a number of leading edge development 
projects designed to advance the computing state-of-the-art. Each of 



4-j- those programs (such as the 701 , SAGE, NORC, STRETCH and various 



projects for NSA*) was a response to the needs and demands of users 
(predominantly government agencies) who required computing capabilities 
beyond the most advanced then available. All of the projects advanced 
the computing state-of-the-art and, in so doing, substantially 
benefited computer users and helped serve the nation. In addition 
they proved extremely valuable to IBM by serving as training grounds 
for future IBM managers and engineers and proving grounds for impor- 
tant new concepts that were incorporated into subsequent IBM computer 
if products (See pp. 68-76, 126-35, above.) With the first STRETCH 
computers commencing shipments in 1961, IBM began work on its next 
"super computer". (DX 4775.) 

The SPREAD Report contemplated the development of a "very 
large processor" beyond that which could easily be made compatible 
with the rest of the line. (Brooks, Tr. 22713-14; Knaplund, Tr. 90477- 
78, 90517; DX 1404A, p. 16 (App. A. to JX 38).) Work on the "high end" 
was under way even as the SPREAD Committee was meeting. As Brooks 



17' 
13 

»i 

i testified: "at any point in time there was somebody working on a 



22 
23 
24. 



machine beyond the fastest one we had; in any project there should be 
somebody looking for a successor to it".' (Brooks, Tr. 22844.) 



| * The NSA projects are discussed at length in the classified NSA 
25 1 stipulation which is DX 3420A, at «Mf 79-86, 333-369. 

i 

':! -401- 



C i 



"In January 1961, a general review was made in IBM of the 
state-of-the-art in components and organizational improvements , with 
the goal of making a successor to STRETCH. ... In August 1961, the 
program was designated Project X (ten times STRETCH)." (JX 10, f 5.) 
A general timetable for development was decided upon, and deliveries 
projected for 1966 or 1967. Responsibility for Project X was given 
to the Data Systems Division in October 1961. Development of the 
Project X computer, which was later redesignated "Project 604" and 
which ultimately became the 360/90 program,* proceeded throughout 
1961-1963. (JX 10, if 5.) The Model 90 program was an effort to 
"push technology" and build "the most powerful computer" possible at 
the time. (Knaplund, Tr. 90571-72; PX 1034; PX 1036; PX 1041.) 

The impetus for the Model 90 program was much the same as 
the impetus for IBM's earlier efforts to "stretch" the state-of-the- 
art. Beginning in 1961 and carrying through the Model 90 announce- 
ments in 1964 and 1965, an increasing number of "leading-edge" 
- i customers requiring advanced solutions to complex computing problems 
began "pressing" IBM for systems with higher performance than I3M 



1 
y 

3 



I 

3 



3 :j 

a 



then had available. (Wright, Tr. 12903-94; JX 10, 1f1f 4, 9; PX 1061.)** 



J Not surprisingly, as it had in the 1950s, a good deal of this pressure 



■i 



•1 

2:1 * The Model 90 program consisted of the System/360 Models 2092 I, 
j 2092 J, 2091 and 2095. (JX 10, II 1.) 

3 j 

| ** Such customers included the Atomic Energy Commission (AEC) 

'A i facilities, the Weather Bureau, various universities, and the National 

! Security Agency (NSA) , as well as private research organizations. 

15 j (JX 10,"«f«f 4, 7, 9, 13.) 

'j -402- 



L 
2 

Z. 

4* 

§' 
S 

r 
a 

la 

a 

VL 

12 

la 
IT 
IS 



came from the Federal governments* 

"'Because of Its need for more and more computer 
capability, the government has encouraged EDP suppliers to 
advance the state-of-the-art. For example, in the past 20 
years, the total, computing power within AEC supported 
facilities has on the average almost doubled each year. 
AEC has encouraged EDP manufacturers to advance the state- 
of-the-art because of its requirements for advanced 
computers. Throughout its history, AEC has acquired some 
of the most advanced computers available." (JX 10, fl 15; 
see Knaplund, Tr. 90920-21; PX 1061; Plaintiff's Admissions, 
Set IV, U 37.0, 53.0-.6, 82. Or DX 7518, Mount, pp. 63-64.) 

In the climate of the early 1960s, such demands were not taken lightly. 

As Dr. Robinson, IBM's Director of Scientific Computing testified: 

"At that time in history, the President of the United 
States and the people at large had dedicated themselves 
towards a substantially higher level of scientific and 
engineering and technological achievement than the country 
had experienced prior to that time due to a variety of 
considerations, including the Russian success in areas of 
technology and science, and a national goal had been stated 
relative to the need for the country to achieve great leaps 
forward in various areas of science and technology." 
(Tr. 23049.) 



Knaplund testified that in August-September of 1963 "IBM top manage- 
ment was deeply concerned that IBM's efforts had not yet developed a 
competitive offering for a number of very large and influential 
users , especially the federal government laboratories for atomic 
is ;i energy research, weapons development, space exploration and weather 

20 j research, and defense contractors to the government".** (Knaplund, 

21 ■■! • 



i 
22. ;| * According to Knaplund, government qsers and contractors were 

; j "right in the forefront" of customers who had the "largest and most 
23 j demanding computational requirements and therefore needed the most 

'powerful computing equipment". (Tr. 90921.) 
24. 1 

■j ** "Mr. T. J. Watson, Jr. , and others expressed concern that IBM 
25 j was not responding adequately to the needs of the United States 

■i 

! -403- 



L 

i 

? 

.4 



Tr. 90518-19; see Wright, Tr. 12897, 12893-94.) 

Development of advanced, state-of-the-art computers was not 
only in the. nation's interest, but in IBM's self-interest as well: 

First, as demand for such capabilities increased., so too 

did the potential business opportunity in meeting those demands. 

Thus, in August 1963, T. V.. Learson wrote: 

"I am informed that a machine 10 times 7090 has a market 
of some 53 machines. If the market is anywhere near this 
number we will be committing a very serious crime in not 
moving Project X . . . at a more rapid pace". (PX 1040; 
see also Norris, Tr.. 5617; Wright, Tr. 12893-94; Brooks, 
Tr. 22718-19; Knaplund, Tr. 90920.) 

Second , there was promotional value in being able to offer 

the world's most powerful computing capabilities to solve the problems 

of highly advanced users. As Wright testified: 



"[I]f you could take one of those leading edge 
customers, or prestige, if you want to use that term, and 
get him to use a data processing system to solve a new 
problem that other people had not yet solved, then 
~ ( j generally many other people would follow his leadership 
m ?! and use the data processing system in a similar way to 
■! solve a similar problem." (Tr. 12899-900; see also Dunwell, 
Tr. 85840; PX 1041; PX 1082; PX 1160.) 

5 



3 : 






Third, the opportunity to work on projects at the techno- 
logical leading edge of the industry offered a powerful incentive for 



II | Government for advanced EDP systems in connection with the Government's j 

j high priority defense and related programs". Thereafter, he ordered j 

22 :• that IBM inquire of government users directly to make certain that i 
.j their needs were being taken into account in IBM's "super computer" j 

23 j (Model 90) development, and ordered acceleration of development j 

efforts on a more powerful computer than even the Project X computer. j 

It \ (Knaplund, Tr. 90519-20.) t 



-404- 



? 

8 
9 

10 

LI 



L the best young talent to come to work for companies who undertook 

2 •{ such projects. These projects therefore served as important training 

3 •:, grounds for future employees. IBM's experience on SAGE and STRETCH 
4- j had provided ample proof of the benefits to be gained in that respect. 
3 if (Dunwell, Tr. 85549-50; Crago, Tr. 85979-80.) 

Fourth , "super machine" development held the promise of 
substantial future value which would be realized through the incor- 
poration of new learning in later products. (Eckert, Tr. 836-37; 
Lacey, Tr. 6657; DX 13526, Forrest, pp. 106-07.) This benefit, 
although quite tangible, was difficult to quantify in advance ♦ IBM's 
experience on STRETCH had shown that, although high technology pro- 
y* |j jects might lose money when all the costs of research were allocated 
»-• | to them,* they could still turn out to be very profitable in terms of 
•* | "technological fallout". (Gibson, Tr. 22593; Case, Tr. 73606-08; 

•- 1 Dunwell, Tr. 85791-94, 85880-82; Hurd, Tr. 86595-98.)** "Many of the 

*\ 

_„ '■{ technological developments made in the STRETCH program were of sig- 

Ics ;{ 

. | nificant benefit to other IBM programs." (JX 10, 1[ 3; Case, Tr. 

j 73606-08; Dunwell, Tr. 85538-49; Hurd, Tr. 86592-93; E. Bloch, Tr. 

!| 91485-89; DX 3171; DX 8923.)/ Thus, T. J. Watson, Jr., writing to IBM 

13 ;i 



fKf^ ill ■ I i I 

I * One of the problems in trying to evaluate the profitability of a 

21 i program like STRETCH is that the value of technological fallout was 
■j not credited to the program nor were the costs allocated to the 

22 "\ benefited products under IBM's internal method of cost allocation. 
| (Knaplund, Tr. 90526-27; JX 10, 113.) That value had to be taken 

23 j into account in deciding whether to embark on a like program. 
I 

2X | ** That view was held outside of IBM as well. (See DX 5423, 
■j Smagorinsky, o. 94; DX 13526, Forrest, pp. 106-07.) 

25 -i 

!• / In a letter written to Thomas J. Watson, Jr., on April 8, 1964, 
| Stephen Dunwell, who had been Manager of Project STRETCH, called 
.| System/360 the "image of STRETCH" because of all the 360 features 
| which first appeared in STRETCH. (DX 3171.) 

j -405- 



President A. L. Williams in May 1965, stated: 

"Although four or five years ago there was some doubt 
as to whether or not we should continue to try to lead 
in this area because of expense and other considerations , 
at some point between two and three years ago, it became 
evident that the fallout from the building of such large- 
scale machines was so great as to justify their continu- 
ance at almost any cost. Therefore, for the past two 
years, under Vin Learson and Dick Watson, this subject has 
had the highest priority, at least in the upper areas of 
the management of the corporation." (PX 1469.) 

There were many others within IBM who felt the same way. For example, 

Dr. Gibson, then IBM Vice President and Group Executive, testified 

that one reason for undertaking the Model 90 program had been that 

"the designing, building and operation of such an advanced 
computing system had, in the past, and was believed would 
continue to provide very valuable experience in programming, 
in architecture, in reliability and in technology". (Tr. 
22644.) 

And Dr. DeCarlo, IBM Director of Systems Research and Development, 

£j[ wrote in June 1964 concerning the Model 90 program: 

"We can be intuitively sure that the technological benefits 
which will flow from this commitment will filter through the 
5 'j rest of the product line. Surely there can be no doubt the 

STRETCH program spawned highly successful financial programs". 
(DX 7692, p. 3; see also McCarter, Tr. 88408; DX 1141.) 



j 
3 



Although these reasons for embarking on the Model 90 program 
antedated the announcement of the CDC 6600,** CDC's announcement 



.. j ** The CDC 6600 was publicly announced in July 1962 (JX 10, 11 4) 

: " ■! but was discussed with customers earlier. (Norris , Tr. 5934, 5938; 

.„ 1 DX 308; DX 309; DX 310; DX 13526, Forrest, pp. 191-97, 205-06, 225-30, 

— j 232-42, 245, 504-08, 570-74, 580-81.) 



-406- 



1 

2 

■5 
4~ 

3 



brought the importance of that program home to IBM management with 
greater force. STRETCH had maintained IBM's lead in the large scale, 
advanced computer field. (Dunwell, Tr. 85736, 35741-42; see also PX 
1469.) Within. IBM., the CDC 6600 caused concern about IBM's industry 
leadership in state-of-the-art computing and about the perception of 
IBM's role, by its customers. In August 1963, T. J. Watson, Jr., wrote 



7 || "Last week CDC had a press conference during which they 

officially announced their 6600 system. I understand that 
in the laboratory developing this system there are only 34 
people, 'including the janitor.' . . . 



3 
9 

ia 



12 

13 

14 



"Contrasting this modest effort with our own vast development 
activities, I fail to understand why we have lost our industry 
leadership position by letting someone else offer the world's 
most powerful computer." (PX 1045.) 



The matter of computers having very advanced capabilities 
was a "top priority" among the subjects discussed at the September 5, 
1963, IBM Executive Conference in Jenny Lake, Wyoming. (JX 10, tf 7.) 
These discussions - included "what actions could be taken by IBM to 
.j- jl catch up to and surpass CDC in the area of very high performance 
;j computer systems". (Knaplund, Tr. 90519-20.) IBM Research was 

i! 

instructed by IBM's Chairman 



18 J 
20 ! 



21 ;| 

i 

22 : | 

23 



"to ensure that IBM does have clear leadership in the computer 
field — meaning a computer which is sufficiently far ahead of 
any other computer — that it will maintain that position of 
leadership and prestige for at least three or four years after 
announcement". (PX 1049.)* 



* Watson, himself, wrote one month later: 



.1 "As leader in the industry, I don't see how we can afford any 

24- "} other position than having the most powerful machine on the 

| market. . . . [W]e should promptly commit ourselves to a 

25 I machine of sufficient power so that our leadership will be 

I unquestioned". (PX 1051.) 



-407- 



I 



DSD was instructed to move ahead "as fast as possible" with Project X 
(which was already planned to have twice the capability of the CDC 
6600) and Research was instructed to accelerate its work toward a 
machine with ten times the capability of Project X* (JX 10, If 7; see 
also Knaplund, Tr. 90520; PX 1021; PX 1036; PX 1041.) 



* ;i While there were reasons independent of CDC for undertaking 

r i the Model 90 program, CDC r s growing success spurred IBM to advance 

J ; the pace of the program by increasing the time and resources allocated 

?! to it. (JX 10, 1f 8; PX 1021; PX 1041; PX 1082; PX 1204.) Neverthe- 

2 ; less, the Model 90 was not announced with the rest of the 360 Series 

i 

1 ; on April 7, 1964, because Paul Knaplund (who was responsible for 

2. I bringing before IBM management recommendations concerning the number 

a i| of processors to be announced with System/360) "did not feel that the 

4. I Model 90 had progressed far enough to warrant a general announce- 
ment". (Knaplund, Tr. 90520-21; see also DX 9080.)* The first Model 

5 ;| 90s — the 2092 I and 2092 J — were announced on August 17, 1964.** 



7 : 
3 

:» 

..1 
»i 



!i ) * Customers were informed, however, "that the Model 90 dev«»1nnm*»nt 

"~ j effort was under way. That information was supplied in a footnote to 
y? '; the System 360 announcement". (Knaplund, Tr. 90521; JX 10, «f 1.) 



it \ 

:{ 
55 ! 



** No Model 92s were ever delivered. It was superseded by the 91 
and 95, which had improved memories. (JX 10, *f 31.) 



-408- 



L 
Z 

111 

i 

i 

5- ! 

t 

I 

« I 

7 
3 

2 

10 

11. 

I 

12. i 

i 

13 '■ 

i 

14- 1' 



The Model 2091 was announced in November 1964.* (JX 10, \\ 1.) Each 
of the Model 90 systems delivered to customers performed well and to 
customers' satisfaction and passed acceptance tests imposed by the 
government where such testing was performed- (McCarter, Tr. 88413; 
DX 3162, DX 3167, DX 3224, DX 3266.) 

The first Model 90 computer was delivered nine months late 
because "IBM encountered unexpected, substantial and critical problems 
in the Model 90 circuitry (ACPX) in 1965". These circuit problems 
were "a major reason for the slippages in the delivery of the Model 
90 computers." (JX 10, 1f 30; see E. Bloch, Tr. 91940-43; JX 10, 
% 33.)** The principal problem, known as the "cracked stripe problem" 



* Product Test non-supported these announcements because it could 
not perform its standard type "announcement testing". (JX 10, 1M( 17, 
21; PX 1177.) McCarter explained: 



"To do this [to undertake the Model 90 program] it was 
necessary to work closely with customers to understand 
their needs. This requirement for customer involvement 
•* !• meant that public disclosure of intention and negotiation 
^ ; with individual customers had to precede the development 
17 ;| of a product to a level where Product Test could conduct 

,[• announcement testing. Hence, the Model 90 program was 
IS ;! not susceptible to the kind of product testing applied to 
other parts of System 360." (McCarter, Tr. 88409.) 



* «• t 



TO ! 



j According to Knaplund, because of the "very advanced technological i 

2Q j nature of the program" IBM management placed "primary reliance" on the j 

I judgments of IBM's top technical people in proceeding with the j 

21 ;! announcements. (Knaplund, Tr. 90523-24.) After deciding not to j 
j announce in April 1964 but before deciding to do so in August 1964, j 

22 "» IBM management had already received information from the National i 
j Security Agency that the (ACPX) ASLT circuitry on which the Model 90s j 

22 \ depended was feasible. (See the classified NSA Stipulation, DX 3420A, I 

j u 387-415, especially <Mf 403, 411-415.) | 

24 1 S 
j ** Advanced computers have frequently been delayed because of j 

25 ! unforeseen problems. (JX 10, II 34.) 

1 ! 

S i 

1 ! 

] -409- ; 



could not have been foreseen because it appeared only when a 
sufficiently large number of components had been put together in an 
operating machine. (Gibson, Tr. 22640-41; E. Bloch, Tr. 91940.) 
• That problem was discovered much earlier than it would otherwise have 
been because of the high current densities in the Model 90 circuits. 
As a result IBM was able to correct the problem on the rest of the 
360 line before most had been built and to inform the rest of the 
industry about the problem- before they ran into similar difficulties. 
(Case, Tr. 73594-95.) 

Discovery and solution of the cracked stripe problem was an 
example of the kind of technological fallout expected from the Model 
90 program. As that program proceeded, additional fallout resulted 
from developments in 

- thin film technology (Gomory, Tr. 98273-75; JX 10, 
fl 32; DX 3164) ; 

!j - monolithic circuitry (Case, Tr. 73593; JX 10, «f 32; 
DX 3164); 
transistor technology (JX 10, 1f 32); 

- packaging technology (Case, Tr. 73593; JX 10, 11 32; DX 3164); 
jl - interconnection technology (DX 3164) ; 

: ! - memory technology (PX 3050; DX 3164); and 

- :! - 

i - machine organization (Case, Tr. 73593; PX 3050) . 

I :! 

| Although the anticipated technological fallout from the 

3 i 

| Model 90 program was realized, the 90 series did not fare well 

- S 

;! competitively. Only 15 Model 91s were manufactured (four for internal 



-410- 






L 
2 
3 

4- 

? 
a 
7 
3 
9 
10 

11 ! 

12 I 

13 i 



IS 



17 i 

IS ; 



use) and two Model 95s were manufactured "specially for NASA".* 
(JX 10, If 35.) 



19 i * By contrast, "CDC manufactured 94 Model 6600/6700 computers and 
:| 121 additional 6000 Series computers". (JX 10, % \ 36.) Mr. Norris, 

2G | Chairman of CDC, called the 6600 "particularly" successful. (Tr. 
i 5849-51.) And in 1969, CDC Vice President, J. W. Lacey, speaking to 

21 ■) a CDC graduate orientation class, said that CDC was widely recognized 
■} to have "a world-wide leading position in large computers" — an area 

22 ■< which CDC was able to "dominate" after delivery of the 6600 in 1964. 

1 (DX 438, p. 7.) According to an IBM offer of proof (DX 1185, pp. 3-4. 

23 j CDC's revenues and gross profits between 1964 and 1972 from the sale 

and lease of 6600s exceeded CDC's targets, and although DX 1185 was 
2i '\ not received in evidence, we rely on it because it is consistent with 
■I the other evidence more fully set forth below in the CDC history. 

25 j 

i 
i 

I -411- 



g* The 360/44 . As discussed above, one of the risks in 
providing a line of computers like System/360 intended to do all 
applications equally well was that, for some applications, at least, 
the machines of the family would be, less suitable for some customers 
'. than competitive machines, optimized in their design for such appli- 
cations. Additionally, there was a risk that not all customers 
j| would be willing to accept the "overhead" associated with System/360 f s 
:| highly functional systems software — that some number would attempt 
to locate alternatives with less function and better price/perfor- 
mance. * 

For some (certainly not all) users in such areas, this 
turned out to be true. Knaplund testified that: 

"In the months following the System 360 announcement, 
marketing personnel began to report that, although many users 
found the System 360 products adequate for data acquisition and 
data reduction, some felt that a general purpose processor more 
tailored to those specific applications would be required. The 
Data Processing Division urgently requested that the Product 
Group undertake development of a system to meet these needs." 
(Tr. 90539.) 



The need for a competitive response became increasingly 
;j apparent during the latter part of 1964 and into 1965. Learson 

•| wrote to Watson in December 1964 concerning the acceptance of the 

.1 
'\ Models 40 and 50 in the "Intermediate Scientific Area": "Our position 

i 
. 'here since announcement in April, 1964 is that we have won 44, lost 

■J 
Li 44, and have 172 doubtful situations. CDC and SDS have a total of 
i 



L | * In this context price/performance means strict throughput per 

j dollar leaving aside questions of function, reliability and service 

r i 

• j 

■| 

] -412- 



L 
Z 

J 

J 
5 



five machines which out -price, out-perform us by a good margin". 
(PX 1288, p. 2.) This was just one of many cries for a competitive 
answer. (See PX 3615; PX 3630; PX 3566; PX 1439A.) 

In August- 1964, DSD began a program, (called the "42S") 
whose objective was to develop a processor "within the general 
architecture of the System 360 family" but with better price/perfor- 



j lj mance than the Models 40 and 50 for "data acquisition, data reduction 



and certain scientific calculations."* (Knaplund, Tr. 90539.) That 
program culminated in the announcement of the 360/44 in August 1965. 
(PX 158 9A.) 

The Announcement Letter described the Model 44 as "a power- 
ful computer . . . designed specifically for the small to medium- 
sized scientific user * . . ideally suited for customers and prospects 
who want raw binary speed and high throughput to solve a wide range 
of scientific problems, including high speed data acquisition jobs". 
(PX 1589A, p. 1; see also Knaplund, Tr. 90539.) To reduce costs and 
achieve the "raw binary speed and high throughput" needed for this 
"lean, hard system", some sacrifice in compatibility with the rest of 



3 

1Q 

Li ; 

12 

12 

L£ 
15 
IS 

it ■; 

IS 

A 

j * In April 196 5, Knaplund wrote that, "The performance needed [in 

20 i the Model 44] approaches the Model 50. The system price required is 
I close to that of the Model 30". He went on to say: 

21 •! 

j "Wherever possible within the framework of our main thrust 

22- :! price/performance curve ... we must and will bend every 

'! effort to preserve complete compatibility for marketing, as well 

23 | as programming reasons. But when an anomalous performer is 

| required, we must be prepared at all times to offer lean, hard 

24 'j systems with slight incompatibilities, if these incompatibilities 
j help mitigate impact and/or cost. 

1 "Such is the case with the Model 44 . . . ." (PX 1439A.) 

i 

:j -413- 



System/360 had to be made.* (Knaplund, Tr. 90540; PX 1439A; PX 1538; 
' PX 1589A, p. 1.) The required cost savings were achieved "by 
; eliminating read-only storage through the utilization of hard- wired 
; logic for the interpretation and execution of stored program ins true- 
jtions, by reducing the number of instructions directly executed by 

i 
i 

I this hard-wired logic, by simplifying the checking logic and by taking 
advantage of lower component costs". The required performance 
increase was achieved "by using hard-wired logic in place of read- 
only storage and by including within the processor a single disk 
storage device known as RAMKIT for program residence." (Knaplund, Tr. 
90540.) 

Apart from its inability to execute the complete System/360 
instruction repertoire,** the 360 Model 44 was "basically the same" 
as the other 360 processors. (PX 1541, p. 6; see also PX 1589A. ) A 
"New Product Programs Status Report on the Model 44 Program", dated 

i one month prior to the announcement, even indicated that the 44 would 



5 

7 1 

i 

i 

5! 

9 





* The IBM Product Group Policy for Processor Architecture (released 
July 30, 1964) envisaged the need for deviations from compatibility in 
order to "keep pace with systems technology and market requirements". 
Exceptions from the rule of compatibility were permitted only to 
achieve cost or performance improvements greater than 10%. (DX 9036.) 
The improvement anticipated in this instance "substantially exceeded 
10%" and was therefore "consistent" with the Policy. (Knaplund, 
Tr. 90540.) 



2 
3 

4 J 
i 

5 



** Even this difference could be eliminated, albeit at some sacri- 
fice in throughput. At announcement, IBM offered as an RPQ an extendeji 
instruction set package (implemented primarily by software) which 
eanbled the Model 44 to execute the "full range of System 360 
instructions". (Knaplund, Tr. 90541; PX 1589A, p. 1.) In 1968, an 
improved version of this feature was provided. According to the 
announcement letter this "Commercial Feature" offered approximately 
a 20% improvement in internal performance compared to the prior 
RPQ. (PX 3563A. ) 

-414- 



1 1 

2 ; 
2 \ 

5 
6 

7 
S 
3 

10 
II 
12 
12 

14 

IS 
17 

18 » 



be manufactured in the same facility as the Model 40 and that schedule j 
restraints would require the substitution of Model 44s for production 
of Model 40s "on a one-forgone basis." (PX 1541, pp. 4, 6.) This sub- 
stitution never happened because additional manufacturing capacity 
sufficient to meet the demand for both Model 40s and 44s was authorized 
prior to the Model 44 announcement. (Knaplund, Tr. 90542? DX 1154; 
DX 1155, p. 2.) 

The Model 44 was not particularly successful. It failed by 
a wide margin to meet the level of acceptance forecasted at the time 
of its announcement. (PX 2163A, p. 4; PX 2419, pp. 6-8.) At least to 
some in IBM, it appeared that this was because IBM had learned to meet 
customer needs generally, but had not successfully learned to specialize 
within that talent. Thus, Opel, who at the time was IBM Vice President 
and Assistant Group Executive, Plans and Controls, wrote in 1967: 
"Why has this happened? Are we unable to plan, build, and market a 
specialized machine?" (PX 1974.) And again, in August of that year: 
j "It seems to me that when we specify a product for a limited market,' it 



i fails. Perhaps this is due to the way we sell or, perhaps it is due 



20 



to the realities of market acceptance. I'm not sure which." (PX 2099; 
see also PX 3555. ) 

1 In part, however, the Model 44 was unsuccessful because it 

21 j 

j was relatively quickly outperformed by later systems of competitors. 

LL ..» 

t 3y the end of 1967, at least some in IBM believed that "hardware per- 

ii j 

I formance was excellent at announcement time, but recent competitive 

2d 1 

j announcements have now bypassed the Model 44". (PX 2125, p. 48.) That 

25 j 

I -415- 



1 

2 

3 
4 

e ; 

7 
8 
9 
LO 
LI 
12 
13 
L4 

15 

ii 
16 Ii 

»i 

ji 

18 I 

19 j 

20 | 
21 
22 

23 i 

i| 

24 || 

ii 

25 Ii 



situation continued to worsen so that by 1970 one group in the 
company wrote: "As a result of being consistently outperformed by 

:the XDS Sigma 5, PDP 10 or CDC 3300, the Model 44 is seldom proposed," 

j (PX 2567, p. 93; see PX 2871Av) 



-416- 



z 

3 

4- 

? 
7 

a 

10 

LI 
12 
12 

14 

15 
17 



h • The 360/67. The SPREAD Report called for the New- 
Product Line to be communications -oriented, multiprogramming sys- 
tems that would be capable of performing time sharing.* (Brooks, 
• Tr. 2285*-60; DX 140-4A r pp. 12, 18, 19, 24, 26, 33 (App. A to JX 38).) 
That objective was met, and 360 as announced included time-sharing 
capability.** (Brooks, Tr. 22859-60; Knaplund, Tr. 90532-33.) 
However, a number of highly sophisticated customers with advanced 
requirements rejected 360 f s time-sharing approach and demanded 



* Time sharing refers to "the use of a computer by many people at 
once with each user having the illusion that he is the sole user of 
the computer". (Perlis, Tr. 1862-63; JX 1, p. 115.) 

** This was not the beginning of IBM's involvement in time sharing. 
IBM participated in a number of time-sharing development efforts 
before System/360 was ever announced. For example: 

— Both SAGE and SABRE were rudimentary time-sharing systems. 
(Wright, Tr. 13664-65; Crago, Tr. 85975-76.) 

In 1960-61, Dr. Gerrit Blauuw of IBM designed a "dynamic 
address translation unit" which was a predecessor for the 
dynamic relocation hardware (Blauuw Box) used in the Model 
67. (Brooks, Tr. 22866; Wright, Tr. 13332.) 



j — In the "early sixties", IBM developed a system that would 

IS ;} execute FORTRAN programs interactively and edit them — one j 

: j of the "important efforts" in adapting a batch processing ! 

13 | language to time sharing. (Perlis, Tr. 2042-43.) ! 

J ! 

20 j — In 1963 MIT, working with I3M, implemented CTSS (which i 

} Perlis called "the first example ... of a general purpose j 

21 i| - time sharing system") on IBM 7090 series systems. (Perlis. j 
I Tr. 1881; see also Brooks, Tr. 22739-40; Morse, Tr. 30986.) 

22 : | CTSS was described by Perlis as, a "creative masterpiece". | 
j (PX 299.) ! 

22 :} i 
[Additional time-sharing work, including work on the design of reloca- 

21 I tion hardware, was ongoing in various IBM research labs. (See Wright, 

[Tr. 13325-28; Knaolund, Tr. 90533; DX 4823.) i 

IS ! ■ ; 

j i 

i : 

■i ! 

■j -417- 



» ! 

M 






.4- 



time- sharing facilities not available with System/360, specifically 
dynamic relocation hardware.* 

In. early 1964, Project MAC*-* at MIT sought proposals for 
the development of "an extremely advanced timesharing system", (Weil,. 
Tr^ 7108.) IBM. bid a multicomputer configuration of a System/360 
Model 50; CDC bid a 6600; RCA bid its 3301; GE bid a 635; and Univac 
bid "a complex multiprocessor system" then being designed for a 
classified military weapons system. Digital Equipment Corporation 
bid "a multiprocessor version of its PDP 6 computer" and was "in 
among the finishers". The winner was General Electric and, in addi- 
tion, "a $1 million PDP-6 was purchased by MAC as a peripheral pro- 
cessor". (PX 2961, pp. 1, 3, emphasis omitted.) GE won with a 
- II "modified" version of the 635 and "proposed working jointly with [MIT] 
in the development of the software that would reside on that hardware". 
(Weil, Tr. 7111-12 J MIT "had determined that System 360 would not 



^ It satisfy their needs and that they would accept only a system incor- 
• porating some form of dynamic relocation hardware". (Knaplund, 
I Tr. 90533.) 

.3 ] 

| * Dynamic relocation hardware provided a "means for interrupting a 
^ :j program at an arbitrary point, moving it out of core, proceeding with 

j the interruption, bringing the interrupted program back into memory . 
II ! at a new location, and starting it again". (PX 1194A, p. 3; see also 

] Weil, Tr. 7287-88; Wright, Tr. 13331-32; Knaplund, Tr. 90532-33.) 

iZ '.! 

■j ** project MAC was an advanced research project in time sharing 
2 t funded by the Advanced Research Projects Agency (ARPA) of the 
Department of Defense. (Wright, Tr. 13288-89; Weil, Tr. 7111; 
Zl 1 DX 5613, pp. 2-3.) 

i 

-418- 



L 
2 
1 
4- 

5 

e 

7 

a 

9- 
10 

11 

12 
13 

14-! 
15 
IS ; 
17 ! 
IS ! 



Shortly thereafter, Bell Labs also ordered a time-sharing 
system from GE. (Weil, Tr. 7116-17.) Rooney testified that, after 
the MXT and Bell Labs orders, within IBM. "there was a great deal of 
talk about the need for such a system in our line". (Tr. 11747.) 

In. mid-August 1964, IBM formed the Time Shared Task Force 

"to develop an IBM plan for time shared systems . . . because 
of the loss of the MAC account at MIT and other critical 
customer situations in the area of real time, time shared 
systems requirements" . (PX 3502, p. 1.) 

The task force was comprised of individuals in IBM "most knowledge- 
able" about remote computing and time sharing, who in turn scheduled 
meetings with a number of the leading experts in the field such as 
Professor F. J, Corbato (of Project MAC), Dr. J. C. R. Licklider 
(of ARPA) , Mr. J. Schwartz (of SDC) and Dr. B. Galler (of the 
University of Michigan). (PX 3502, pp. 3-7.) 

In early September, Nat Rochester, a member of the Task 
Force (PX 3502, p. 2) wrote to C. H. Reynolds, the Chairman: 



"System/3 60 has been almost universally rejected by the 
leading time sharing investigators. Time sharing systems 
are likely to render obsolete systems that are not based on 
time sharing. Therefore, there is a legitimate worry that 
System/3 60 may not be a resounding success unless proper 
£ Q ;j steps are taken." (PX 1194A, p. 1.) 



?q : j He stated that "the commonest reason the customers give for rejecting 
71 j System/360 for time-sharing is that there is not adequate hardware 

22 '• support for dynamic relocation", even though "dynamic relocation is 

23 :| not actually required for time sharing". ( Id. , p. 3.)* IBM was being 



24 i! 



25 j 



* This was also the view of Fred Brooks, chief architect of 
System/3 60, who held the opinion that dynamic relocation hardware 
j "was unnecessary for time-sharing or any other purpose". (Brooks, 
I Tr. 22743.) That is why such hardware was omitted from System/360. 
J (See also Knaplund, Tr/ 90532-33 . ) 

j -419- 



19 i 

20 | 



told that 

"customers want dynamic relocation. It may be unnecessary and 
undesirable but we have not yet proved that this is so. The 
technical situation is very unclear and is changing rapidly on 
a month by month basis as technology advances." ( Id. , p. 4.) 

Reviewing the "rejection of System/360" by those desiring time 

sharing, Rochester concluded: 

"There is much more at stake than these few prestige accounts. 
What is at stake is essentially all computing business, scienti- 
fic and commercial .... we may find eventually that many of 
the best programmers will refuse to work at an installation 
that does not offer timesharing or offers inferior time sharing ►" 
(PX 119 4A, pp. 2-3.) 

He recommended that IBM "proceed with the design, construction and 
release of an advanced timesharing system," and that the work be done 
in public "so as to benefit from external criticism and so as to have 
a favorable sales effect". ( Id. , p. 1.) 

Two days later, the Research group of the Task Force reached 
the same conclusion: "System/360 has been rejected or is about to be 
rejected by many of the important large-scale scientific users who are 
pioneering novel ways of using computers such as the 'computer util- 
ity* . This has been accompanied by a shift of attention to compe- 
titive equipment like the GE 635." (PX 2811, p. 1.) They also 
believed there was "a great deal more at stake": 



: { "The earlier concept of 'time-sharing' has now naturally led 

2 ~ !{ to the 'computing utility' concept. This means that computing 

;! capacity should be available right at the working place of the 

22 J computer user by means of a terminal linked to a powerful 

:l central computer. . . . There is a very strong probability 

" j that the 'computing utility' will be the way of all scientific j 

j computing in a few years, and a good possibility that it will j 

2^ ; capture a substantial part of the commercial market as well. j 

if IBM cannot afford to overlook a development of this scope. j 

25 ; We, are currently in danger of losing all contact with the ; 
leading developers of this concept." ( Id. , pp. 2-3.) j 

i 
] 

-420- '! 



L 
2 

Z 
S 

3 

10 

11 
12 

13 

14. 
15 

16 
17 
13 



The Research group recommended that "if IBM is to keep its 
present competitive position in the marketplace", IBM must 

(a) * 'immediately undertake a long range study of the • 
'computer utility* concept"; 

(b) "undertake an immediate all-out educational effort 
to explain the capabilities of System/360 and its operating 
system for the multiprogramming applications involved in most 
teleprocessing and communications -oriented systems (where 
dynamic relocation is not required)"; 

(c) "undertake a project with the objective of offering 
and supporting a complex 'utility type' system requiring 
multiprogramming, multiprocessing and time-sharing with 
System/3 60"; and 

(d) "to make this intent clear", announce "a multiplexor 
channel and hardware- aided dynamic core relocation capability 
at once for Models 60, 62 and 70". (PX 2811, pp. 4-5, emphasis 
in original. ) 

It was thought that only by implementing these recommendations would 



£9 1 IBM be able to retain its "position of leadership which threatens 

20 j to slip from us as a result of the independent development of the 

21 ij utility concept to which we have only belatedly directed our atten- 
22'! tion". (PX 2811, p. 7.) 

23 ! In mid-September 1964, IBM's Scientific Computing Department 

■I 
24, 'i reported on "remote scientific computing" to the Task Force: 

j 
25 1 "There exists in the market place a set of key leader 

"""* j accounts representative of the scientific market segment. 

;l 
i 

1 -421- 



L 
2 
3 



3 

i 
7 I 

3 ; 

s i 

LQ I 

i 
LI I 



13 \ 
14; 
15 ! 



These accounts are invariably the innovative and exoerimentally- 
oriented accounts* They are the industry's spokesmen on the 
advanced state of the art in computing. They materially 
affect computer acquisition decisions in a variety of smaller 
establishments — both scientific and commercially oriented. . . 



£j "... In general ,• the accounts in the set number over one 

hundred. They consist primarily of AEC computing laboratories, 
large University laboratories, large research laboratories of 
industrial companies, the independent non-profit laboratories, 
and certain aerospace establishments. 



"Today, a subset of this market, led by key university and 
certain closely related laboratories, has taken a fancy to the 
so-called area of remote computing. . . . 



"IBM's posture has been one of silence. In the remote 
scientific area we have been at a severe disadvantage because 
12, I we have not made available sufficient information regarding 

our operating system for 360. It has hot been stated to what 
degree the operating system will support, time-sharing. We have 
also not stated what additional support, if any, will be avail- 
able for time-sharing. 



"Our time-sharing prospects require responses to the 

|| specific functions they have posed as requirements. The 

1$ \ balance of the remote scientific community needs to know our 

:t responses in this regard as well as more detailed information 

17 '■ about operating System/360. 

ig • "Certain accounts have already been lost. A small set of 

:j key accounts are right now in the process of evaluations leading I 

£3 to computer acquisition decisions. For every such case, deci- 

; i sions disadvantageous to IBM appear to be in the offering. In 1 

?g J quantity, such losses do not appear to be large. In quality, j 

■j they will have a tremendous impact upon a very large market j 

?-T •! segment. ... f 

j I 

1 

22 : ; "If we do not respond on the time-sharing requirements in ! 

;j the near future, the time-sharing market will be largely lost I 

?~ :| to GE who has responded to this requirement. A large part of j 

""""' i the balance of the remote scientific market will also be in i 

24 ! i 

25 \ \ 

■ ] 

j ! 

' i 

-422- j 

i 1 '• 

; i 

! t 

i ! 

; i i 

'■ l 

I 



3 
9 

10 
11 
12 



13 



14- 



jeopardy. ..." (PX 2964A, pp. 4-6.)* 

The report foresaw that the competitive threat would not 

be limited to GE: 

"We can expect similar emphasis on time-sharing system 
design from the other competitors. The experience of Burroughs 
with the D 825 and of Remington Rand with the M 490, 1218, and 
other special forms of real-time computers designed primarily 
for the military, have provided them with the experience 
necessary to develop well-honed second generation systems 
designed for general-purpose use. CDC has also had experience 
in the design of real-time systems. Furthermore, the system 
study efforts being conducted by CDC and ITEK at McDonnell 
Aircraft, General Dynamics and Lockheed, in the area of 
computer-aided design, will ultimately result in the announce- 
ment [sic] generally marketable equipment to compete with Alpine 
and, to a broader extent, the 2250 and its successors. The 
Digital Equipment Corporation is actively marketing the PDP-6 
as a time-sharing system at extremely competitive prices. 
Although no real manifestation of intent has been made by 
RCA and Honeywell, the ultimate gravitation of the market 
toward general-purpose time-sharing systems will encourage 
all manufacturers to develop a product and support plan. 

"... The growing emphasis in the scientific and engineer- 
ing market must ultimately effect [sic] the system selection 
process among so-called commercial users. ... 



£3 i 



* As we explain elsewhere, GE was at this time a corporation with 
I corporate-wide annual revenues in excess of $5 billion — a 
-^ | "sleeping giant" to be sure, but one with the resources and techno- 
logical capability to become a major force in the computer industry. 
(See below, pp. 483-90.) 



±z 



-423- 



- a 

"The advent of cost-justif ied, time-sharing business on 

17 ; centrally located systems should have an explosive effect on | 

| the service bureau business. This business is characterized j 

XS ! today by the presence of a great many users located remotely l 

ij from central facility. To some extent, the current business | 

?g ;j in service bureaus is limited by turnaround time. Most service j 

;[ bureau customers who install their own equipment do so because j 

2Q \ of the delays introduced by access to a centralized location j 

~] and service." ( Id. , pp. R28-29.) I 

2i i . i 

i ! 

?? '! . ! 



In sum, the whole market, in all its dimensions, would be affected 

by the need for advanced time-sharing capability.* 

The conclusions- of the Task Force were buttressed by 
■ feedback from the marketplace. ' Por example, Hart testified that 
" through the joint study activity which General Motors Research had 

with IBM, 

"we vigorously provided input to them about what our needs 
were and the importance of doing this, job right, and what 
we believed was the right way to do it. . . . [w]e went 
to meetings and presented our case, and we, I suspect, did 
it loudly and with great conviction. Because, if we were 
going to provide a suitable time sharing environment to 
support our graphic consoles, then we needed to have certain 
capabilities available in order to be able to do that ade- 
quately." (Tr. 80278-79.) 

In late 1964 or early 1965, Dr. Ivan Sutherland, Director 
j of Information Processing Techniques for ARPA, contacted V. 0. Wright 
i of IBM "eight to twelve" times to discuss the topic of time sharing: 

"He spoke words of encouragement, encouragement in the 



* The importance of time sharing to the computer market as a 
i; whole was perceived outside of IBM as well. For example, Project 
■ MAC and GE both believed that the computer field would evolve toward 
: "an information utility". (Weil, Tr. 7116.) Various members of the 
'•| computer field within the ARPA network (such as MIT, Stanford, Stanford 
'i Research Institute, Lincoln Laboratories, SDC, Rand Corporation and 
; i the University of California at Santa Barbara) believed that time 
* sharing "was important" and should be pursued. (Perlis, Tr. 1968-69, 
ij 2043-44, 2054-55.) In 1965, Withington wrote that the "market for 
! time-shared computer systems and applications" was "large and growing", 
j He predicted that, within five years, such systems would represent a 
;j "significant part of the total computer market. In fact . . . the 
[great bulk of the computer market. ..." (PX 4830, p. 14.) 



-424- 



L 
2 

1 

4 

5 

S 

7 | 

3 | 

i 
* I 

la I 

11 1 

12 | 

13 • 

14 



fact that he believed that IBM should pursue development 
of the timesharing concept in products and software as 
a matter of not only great importance to the United 
States government, but also of great importance to 
IBM and he simply encouraged and wanted to be kept 
aware, sort of as an insider, of how things were 
going on the project. 

"[I]t was my understanding that his interests were 
the fact that he believed providing timesharing facilities 
to the Department of Defense contractors in design of new 
weapon systems, and use in other things, including health 
systems and so on, would, in fact, foster the use of compu- 
ters, but more importantly from his standpoint, would assist 
in the solving of problems that these people in their research 
and development activity were confronted with and the use of 
computers would facilitate the solution of those problems at 
a more rapid rate and therefore accelerate the advancement 
of technology. 

"[I]t was clear that he felt that two large companies, 
such as GE and IBM, pursuing developments in time sharing, 
was beneficial to the government, was beneficial to industry 
and, therefore, that he thought that was a good situation." 
(Wright, Tr. 13287-92.)* 

IBM marketing people, too, were "raising an increasing 

amount of clamour, putting an increasing amount of pressure on the 



* Sutherland wrote the following on September 4, 1964, shortly 
after GE had been selected over IBM for Project MAC: 



16 a 
17 

IS . 

;j "Project MAC'S decision in favor of G.E. has generated 

jg j a very healthy spirit of competition between MIT/GE and 

■j IBM. In effect, Project MAC has stated publicly that the 

2Q | IBM product is inadequate and that MIT/G.E. can do better. 

) MIT/G.E. must produce the best system they can in order 

2i t to make good their claim. IBM must expend its best effort 

j to show that its product can serve the needs of time-sharing 

22 i l n fact, IBM has been slow in responding to the needs of 

: interactive computer users; now we can expect IBM to show 

23 1 more interest in this field. Competition between IBM and 

! MIT/G.E. is a good thing; it will stimulate rapid progress 

24, j in the time-shared use of computers. 



-425- 



L j marketing management of IBM" to provide "a product response that 

2 ; would let us be more responsive to our customers ' requirements 

Z\ and to our customers' demand". (Wright, Tr. 12799.) By November 

i 

4- j 1964 Wright in the GEM region and others within IBM became concerned 

5 j. that the time-sharing movement would build "to a great ground swell" , 

i 
5 I "impact" IBM's installed base of equipment and result in "a great 

7 I deal of churning of the installed base, that is, the return of 

3 j products that IBM had installed because of the requirement for a 

g. | new capability in a computing system". (Wright, Tr. 12802-07.) 

i 
IQ \ According to Wright, he and his marketing colleagues were "trying 

i 
H S to make sure that IBM was the leading producer, vendor, for data 

12 [ processing equipment and . . . that IBM did not fall behind". 

12 I ( Tr - 12807.) A response was needed "to the customers who were 

j 

14 ! pushing us very hard to provide a product answer to their recuire- 

i 
75 i ments." (Wright, Tr. 12807.) It was clear that others would pro- 

i 

iS ! vide that response if IBM did not. (Wright, Tr. 12843-45; PX 2964A, 

17 | pp. R28-29.) 

,« j One of the catalysts for such response was the Lincoln 

,q '! Labs Request for Proposal which came in November 1964. (Wright, 

: t 
20 ; i 



2^ ] "ARPA must support Project MAC fully. The MIT personnel 

! responsible for choosing G.E. equipment have made their best 

22 \ technical judgment. They are staking their professional 

j reputations on their choice. In making a decision against 

23 ■! IBM, they have stimulated IBM to new efforts. Were ARPA 
j to reject the MIT decision, Project MAC would suffer a 

- 4 | blow from which it might never recover, and IBM would 
be able to relax." (DX 894.) 



-426- 



Tr. 12813.) At the same time, an RFP was received from the 
University of Michigan for a "central, timesharing facility". 
(DX 895 .) Watts Humphrey, IBM Director of Time-Sharing Systems, 
wrote to Le arson on November 15, " [t]he list of accounts who have 
interest in Time Sharing is growing daily. ... By the end of 
the year, I expect that this number will exceed thirty." (PX 1238A, 
p. 4.) Company prestige, as well as current and future business, 
was on the line. (PX 1191; PX 1246A. ) 

The messages from the field were heard by IBM management. 
Knaplund testified that reports from DPD in late fall of 1964 
revealed that a number of "very influential and highly competent 
users"* agreed with the MIT analysis of System/360 and viewed dynamic 
relocation as being "crucially necessary" to a broad variety of new 
and advanced applications — a feature that would "accelerate and 
improve the efficiency of their internal system development and 



t j 

2 \ 

i 

Z : 
i 

_ ! 

f ! 

a ; 

7; 

1 

3 ; 

9- \ 

i 

10 l 

i 
! 

II ' 
i 

12 j 

13 : 

i 

t 

15 \ 
! 

15 ; 

| * These users included MIT's Lincoln Laboratory, General Motors 
17 ![ Research, the University of Michigan, Carnegie Tech, Bell Labs, 

[ Rand Corporation, Stanford University and Ohio State University. 
IS; (Knaplund, Tr. 90534; PX 2811, p. 1; PX 1194A, pp. 2-3.) IBM 

•| Chairman (then DPD President) Frank Cary testified concerning 
13 ;i the Model 67 and time sharing: 

2C ! "[S]ome of our very, very best customers wanted it. . . . 



". . .1 can just tell you that when customers . . . like 
AT&T and the Federal Government and the universities and General 
Motors Research . . . ask us to respond, we certainly at 
least try to respond to them. And we didn't undertake 
2* i that with anv thought that we weren't going to be able 
J to do it." ~(Tr. 101808-09.) 

25 | 



-427- 



L; programming activities". (Knaplund, Tr. 90534.) Knaplund, Hume, 

Z\ Learson, A. K. Watson and others concluded that "an intensive 

3 ;. effort was urgently required to review the area of time-sharing 

i 

4 | axx< ^ develop a plan for meeting this requirement". (Knaplund, 

t 

5 |. Tr. 90534-35; Cary, Tr. 101808-09; PX 1246A.) 

5 I In November, a group reporting directly to T. V. Learson 

7 | and A. K. Watson was set up under the leadership of Watts Humphrey 

i 

3 i to try to respond to the time-sharing requirement. V. 0. Wright 

g | (who was made Director of Time Sharing Marketing) was called to 

^q ! Learson* s home on the Saturday after Thanksgiving Day in 1964 and 

^ i told to begin work that afternoon. According to Wright, Learson 

£2 r said that "the resources of the company were available to us for 

i 

13 j whatever we needed in order to move this development forward". 

i 

L4 J (Tr. 12793-95, 12814-15; Knaplund, Tr. 90535; PX 1318.) 

I 

25 ' Starting in December 1964, IBM made time-sharing proposals 

i 

tg \ to Lincoln Laboratory and "a limited number of other users in order 

yj | to enhance our ability to learn and understand time sharing" . 

1S : (Wright, Tr. 12842-43.) According to Wright: 



20 



21 :| 



"IBM at that point in time was looking at this whole 
development as sort of a learning vehicle or process, if 
you would. There were a great many things about time 
sharing capability in a computer facility that IBM did 
not understand . . . and we went about it on the basis 
that we wanted to develop a product that would satisfy 



25 ;! Lincoln Laboratory and perhaps a few other selected 
:; customers, and that . . . development, would be used, 
then, as a learning process for IBM to understand what 
really a time sharing system ought to be, what the 
~ A i facilities and capabilities should be, both in hardware 
z * \ and in software." (Tr. 12825-26.) 



23 : 



25 



-428- 



L ; IBM delivered its system to Lincoln Labs "four to five 

2 i months later than had been originally proposed". Although it did 

3; : not have all the functions originally proposed and did not perform 

l 

4^j as rapidly as had been anticipated, Lincoln Labs was able to use 

i. 

5 . J it as a time-sharing system. (Wright, Tr. 12829-33.) Wright 

i . 

$ ; believed, considering the fact that it was "the first of a develop- 
j\ ment program", that Lincoln Labs was "reasonably satisfied with the 
a I product" . They 



9- 
10 

11 

12 J 
13 

14 

15 ! 

16 

17 

IS 

19 



"expressed some dissatisfaction in the beginning, and as 
they continued to work with the product and we continued 
to work with them in the product, they became more satisfied 
and the expression of dissatisfaction was eliminated". 
(Wright, Tr. 12832-33.) 

After the Lincoln Labs proposal there was a "great deal 

of demand" for IBM to propose similar products to others. (Wright, 

Tr. 12842.) Perlis testified that he and others at Carnegie Tech 

pressed IBM to provide "the same kind of time sharing service that 

MAC was developing" and were telling IBM that time sharing was 

"important" and ir that what MIT and General Electric had joined 

together to do was the wave of the future". (Tr. 1963-69.) 

Others in the ARPA community did the same. (Perlis, Tr. 2054.) 



:| IBM selected certain users who were believed to have "the 

20 : 

__ -j capability of using a development system" and agreed to propose to 

21 1 

"a limited number" in order to enhance its time-sharing knowledge. 
(Wright, Tr. 12842-43.) From January 1965 forward, IBM worked 
with a group of customers nicknamed the "inner six "--the University 
of Michigan, Lincoln Labs, Bell Labs, SDC, Carnegie Mellon University 
and General Motors. These institutions were selected to act as 

-429- 



22 
23 
24 
25 



L 
I 

t 

j 

z 

T\ 

3 

a 

1 

2 
2 



"consultant or adviser to the group developing the 67" because 
they were "the most knowledgeable and could make the greatest 
contribution to [IBM.\s] designing a product that would fit the 
requirements of [the] user community". (Wright, Tr. 12905-08; 
see also Hart, Tr. 80293.) 

Although IBM had originally intended to propose to only 
six to eight customers "to enhance [its] experience base in the 
i use of the product"', that number was increased "because of the 
great pressure that built up in demand from users and from the 
IBM marketing organization". By October 1965, 63 proposals had 
been made.* (Wright, Tr. 1284 3.) 

And IBM was still quite concerned that its competitors 
would steal a march: 



^ ;■ ■ " [A] great many users . . . felt that time sharing offered 
"" : them some additional capability that they needed. . . . 



.Si! 



J 



'} 



"In some instances they would contact or write a letter 
to one of the IBM top senior executives. In other instances 
they would talk to their salesmen in their facilities, and 
so on, wanting a proposal, wanting to understand what IBM 
could do to satisfy this requirement. 



■3 :J "And all during this period of time, in general, the 

industry was in a state of agitation because time sharing 
appeared that it might indeed be a new wave of the future 
:j from the standpoint of computing facilities for a company 

-G | or an institution. 



:9 \ 



TT .' 



i2 ;j * in early 1965, IBM received and responded to requests for 
; j proposals from NASA, Lewis and various other government agencies 

i3 .j (including certain national security agencies) . (Wright, Tr. 
j 13316-24; DX 901.) 

j 
15 I -430- 



• • •■ 



2 ! 
3 

*! 

si 
s i 

i 

7\ 

i 

si 

9- 
10 

11 

i 

12 j 

13 ! 

■i 

14 | 

15 ! 
i 

is ! 

;j supervision. One or more people "with technical qualifications 

17 » 

11 examined each such proposal ... to ensure that IBM could provide 

IS •' 

•[ those functions" and there was a Review Committee whose approval 

19 • 

; was required before the proposals were submitted. (Wright, Tr. 

*° ;| 13334-35.) 

21 i! 

;j . 

00 » 

;j * The Model 67 was simply a Model 65 CPU, modified by the addi- 
2- jj tion of a "Blaauw Box" (relocation hardware). (Wright, Tr. 13357; 

j DX 898.) The Model 67 could be run as a Model 65 and "many" Model 67 
~ A ;| users did so by running OS part of the time and TSS the rest of the 

jtime. (3rooks, Tr. 22760; PX 2029, p. 1.) 

ji 

25 j 

-431- 



"[TJhere was clearly .... an understanding that if 
IBM for some reason did not respond to this particular 
requirement of customers' need, ... it was very likely 
that those customers might very well buy such capability 
from somebody else. 



"[T]he significance would be that IBM would lose 
business and that part of the installed base that IBM 
had at. that point in time would disappear." (Wright, 
Tr. 12843-45.) 

In March 1965, IBM announced the System/3 60 Models 64 

and 66 "for limited bidding". (PX 6209.) With the availability 

of improved memory for the Model 65 in April, the Models 64 and 

66 were withdrawn and replaced by the Model 67,* which was also 

released "for limited bidding". (PX 1427.) Wright, who was the 

Director of Time Sharing Marketing from November 1964 until fall 

1965, agreed that "every time sharing system proposal made by the 

IBM Corporation during that time" received his close personal 



L 
2 
3. 
4~ 

3 

7 
3 
9 
.0 



LI 



L3 

14 

15 

• *•< 

±3 

17 
IS 



IBM was "very careful to be sure that all of our customers, 
the people who had orders, knew in fact the status of the program, 
what might be a problem, if it existed at that time, and how we 
were progressing."" Moreover, because the customers involved were 
among the most sophisticated users, they "were able to understand 
the technical problems associated with the development effort" . 
(Wright, Tr.. 13336-37.) The customers "understood that the Model 
67 was a research and development project and that things would 
change as they went along", but they were "willing to compromise 
on some of the things that we said would be included in the product 
and give them up if we could not produce them". (Wright, Tr. 
12881-84, 13359.) 

The Model 67 had its special bid restrictions removed and 
was announced in August 1965 for delivery in 1966, with the TSS 
operating system scheduled for delivery beginning in June 19 67. 
(DX 898, p. 2.)* The problems of developing TSS were substantially 
greater than IBM or the customers had foreseen. (Perlis, Tr. 1981-82; 
Knaplund, Tr. 90538; see also DX 13448.) Wright testified that when 

i 

I he left his job as Director of Time Sharing Marketing in November 



19 ij 
11965, he believed there was "some" risk of slippage in the software, 

20 1 

j but "good progress was being made in the development of TSS" and 



•j 

22 " 

■j * Product Test issued a "formal" non-concurrence with the 

23 } announcement , although it believed the program was "in good shape", 
j The non-concurrence was resolved by management. (Wright, Tr. 

2± U3352-54, 13667-68; Knaplund, Tr. 90536-37; McCarter, Tr. 83416-17.) 
j The difficulties which the 67 eventually experienced were unrelated 
25 I to Product Test's reasons for non-support. (McCarter, Tr. 83418.) 

i 

i 

j -432- 



L ! 

i 

3 ! 



4*j- fact that the degree of automatic operation of the system and 



3 

e 

7! 

3 

9 

10. 
Li 
12 
13 
14-1 
15" 
IS 



"the program would be accomplished ... as it was described at 

the time". (Tr. 13360.) By July 1966, however, the number of lines 

of TSS code had "approximately doubled", largely because of "the 



particularly its ability to protect users from each other and 
from system failures is a great deal more complex than had been 
anticipated". However, the first release was still expected to 
be "relatively solid in terms of schedule". (PX 1826, p.. 2.) 

Problems continued to develop. In August 1966 IBM announced 
a delay of 45 days in the release of the initial TSS package. 
(PX 3471.) Further, in the fall of 1966, shortly after learning 
of performance difficulties with the TSS software, IBM made calls on 
its 360/67 customers to explain the situation and to inform them that 
certain functions were being decommitted and schedules delayed. 
(Wright, Tr. 12876-78, 13363-66; see also DX 897.) Wright testified 
that everybody had been informed and understood that this might occur: 



i "All the customers understood that it was a development type 

17 ;j of a project, it was a development of a system that was to 

t some extent breaking new ground, . . . and everybody under- 

IS | stood that there might be changes. ..." (Tr. 12879, see 

!j also Tr. 13364-65.) 
IS I 

■! Hart testified that General Motors Research was kept fully informed 

20 : i 

: | of the problems that IBM was having with TSS. (Tr. 80294.) 

21 -I 

■j In the meantime, GE was experiencing similar problems. 

22 J 
GE's efforts at Project MAC were aimed at developing a software 



system called MULTICS, which was to be implemented on an advanced 

2A i 

.1 

25 1 

i 

■i 

'! -433- 



version of the 635, called the 645. (Weil, Tr. 7227-28.) GE 

announced the 645 to the public in the fall of 1965, when neither 

the machine nor the software was in existence. (Weil, Tr. 

7232-35 •) Before the end of 1966, GE withdrew the 645 from 

marketing because, it 

"began to realize that what we had on our hands was a 
research project and not a product. ... We were attempt- 
ing to do something that had never been done before, and, 
in. principle, we might end up discovering that it was not 
feasible* As it turned out, it was hard and slow, but 
it was feasible." (Weil, Tr» 7234.) ' 

Weil described the GE 645 as "being in the research project stage" 

until 1969 or 1970. (Tr. 7234-35.) In fact, the GE-MIT MULTICS 

operating system was never delivered by GE; Honeywell, after the 

merger with GE, completed development of the software three years 

* !J behind the original schedule. (Weil, Tr. 7232-33; Wright, Tr. 

L\ 13375-76, 13673-74.) These problems arose because "the partici- 

i 

? pants in the Project MAC effort underestimated the difficulty of 

* ! i successfully developing MULTICS". (Weil, Tr. 7232.) As GE's Weil 
testified: 



!i 



"The technical task that was being attempted was extremely 
J sophisticated and many of the subjects were at the state 

! of the art. as it was then known, and it took a long time j 

j to iron out the details of implementing some of these j 

! important features." (Tr. 7232-33.) j 

i . . j 

I , The 645 was never delivered and Project MAC received, I 

j ! 

2 •; instead, a system designated the "636". (Wright, Tr. 13375-76.) j 

3 1 Rather than providing GE with the "top-of-the-line prestige lustre" •; 

i j 

^ j which had been expected, the 645 provided "very little to General j 

'\i ! 

c 1 I 

~ j ! 

1 -434- i 



I 

2 

3: 

i 

*i 

i 

5"! 

t 

1 

si 
t\ 

i 

3 ; 

i 
9-1 

I 

10 ; 

11 ! 

12 ! 

13 ! 

j 

14-: 

h 

15 ; 



Electric except a drain on its resources". (Weil, Tr, 7236.)* 

IBM did not give up on TSS or the Model 67.** Release 1 
of TSS was made available by IBM in October 1967. (DX 3282A.) By 
April. 1969, IBM had delivered a ,r substantially improved" version of 
TSS which was "considered to be an excellent software programming 
system". (Wright, Tr. 12842, 13375; DX 905; see also Hart, Tr. 
80296-3 00, 81961-63; Cary, Tr. 101809.)/ 

The Model 67 was not widely accepted, and by year end 1970, 
only 52 Model 67s had been installed by customers. (DX 2609B, 
p. 182;// see Cary, Tr. 101809.) However, the experience that IBM 
gained with the Model 67 and TSS proved invaluable. Evans testified 
that when he returned to SDD from FSD in 1969,0 he launched an effort 



* IBM's difficulties with TSS and GE's with MULTICS were hardly 
unique in the industry's development of large operating systems, 



particularly for time sharing. GE also encountered problems with 
its GECOS operating system. (Weil, Tr. 7215-19; see also Withington, 

Tr. 56727-31; below, pp. 501-03.) As we discuss below, so did manv 
lo :! others. (See pp. 479, 568-72.) 



17 ! 

is ; 



** The magnitude of the task was so great, however, that IBM did 
consider withdrawing the Model 67 at one point, (px 1955 (DX 13866).) 



:| / That view was not universally held. Per lis testified that 
13 ;j "TSS is working today", but that it never delivered the "work load" 
;| that Carnegie "expected that it should". (Perlis, Tr. 2118-19.) 

20 : 

.; // We are aware that DX 2609B is not in evidence but we rely on 

2! ; | it for the number of Model 67 installations because it is a sworn 

j statement by an IBM corporate officer based upon IBM's accounting 

22 ■» books and records . 

23 j Evans believed that his being sent to FSD was in some measure 
j a punishment tor failing to have dynamic address translation hard- 

24- j ware incorporated into the design of System/3 60 from the start. 

1 (DX 4740: Evans, Tr. (Telex) 3950.) 
25 j 

=1 -435- 



L 
I 



M ! storage capability became a staple of all 370 systems announced 



to get dynamic address translation hardware, the hardware which is key 
to virtual memory systems,* put into the 370 plan. (PX 2487A, p. R2; 
DX 4740- Evans, Tr. (Telex) 3937-41.) Evans was successful, and virtual: 



i after August 2, 1972. (Gary, Tr. 101809-10.) Moreover, virtual 
memory function was incorporated in 370 "in almost exactly the same 
way as the Model 67". (Case, Tr. 73403, 73612-13, 73578-79; Cary, 
Tr. 101809-10; see also PX 2 4 8-7 A, p. R2; DX. 8066.) Thus, the Model 
67 development produced hardware and software that became important 
elements of IBM's computer systems for the next ten years. 



3 

7 

a 
? 
a 

z 

3 

A- 

** 

J I 

.$ ; 

i 

21 ;j 

j * Virtual memory or virtual storage is a combination of hardware 
22. '} and software which allocates to the machine itself the task of moving 

| data into and out of main storage from auxiliary storage. Virtual 

23 j storage greatly simplifies the programmer's task because it relieves 
! him from the burden of having to make sure that his data will fit 

24 "i into available main memory space at all times . For programming 

I purposes, virtual storage gives auxiliary storage the appearance of 

25 'j being main memory. (DX 4740*- Evans, Tr. (Telex) 3943-54.) 
i 

•i -436- 



14 
15 
15 



1 I 33- Educational Allowances. Universities had played a key 

2 | role in the beginnings of EDP in the 1940s and 1950s (" [t]h.e first 

3 'computers were conceived and. built at universities » . ."(DX 5504, p.. 

4 | 15)), and a close working relationship had arisen, between academicians 

i 

5 !| and EDP manufacturers.*' Also, during the 1950s and 1960s, many 

5 ! f colleges and universities, supported in part by the National Science 

j jj Foundation and other government agencies, greatly expanded their 

Ij 
g j utilization of computers* (Plaintiff's Admissions, Set IT, If 641.2.)** 

g ijThe number of campus computing centers grew from 40 in 1957 to 400 in 

in !| 1964 (DX 5504, p. 15), and, as the Rosser Report (DX 5504)/ acknowl- 

ii j 

T<? ! * For example, as noted earlier, Eckert and Mauchly, the developers 
i2 iiof ENIAC, did their early work at the Moore School of the University 
13 ij of Pennsylvania. (Eckert, Tr. 712-15.) John von Neumann, whose 
Ij papers contributed to the development of the modern stored program 
ii concept, was a member of the Institute of Advanced Study at Princeton 
land later a consultant to IBM. (Hurd, Tr. 85614, 86599-600.) Herman 
'iGoldstine, one of von Neumann's closest collaborators, joined IBM 
jjaround 1958 (Gomory, Tr. 98154), and became IBM's Director of Math- 
ematical Science in IBM's Research Division. (JX 5, p. 57.) In more 
'; recent times, Phillip McC. Morse, Director of MIT's Computation 
:; Center, is a member of CDC ' s Board of Directors, and Harold Brown, 
17 .(President of the California Institute of Technology,, is a member of 

'i IBM's Board of Directors. (PX 5779, p. 33; Morse, Tr. 30961.) 

13 ; ; 

** As early as 1956, the Atomic Energy Commission was giving grants 

19 iito universities in order to support the usp of computers. (DX 5424, 
= Pasta, pp. 11-13.) By 1963, at least eight government agencies 

20 ^contributed to -ne support of computers in colleges and universities: 
.National Science Foundation, National Institutes of Health, Atomic 

21 lEnergy Commission, Advanced Research Projects Agency, NASA, Air Force 
Office of Scientific Research, U.S. Army Research Office and Office of 

22 Naval Research. (DX 5504, p. 43.) 

23 / The Rosser Report, published in 1966, was the work of an ad hoc 
committee, the Committee on Uses of Computers, appointed by the 

24 National Academy of Sciences. J. Barclay Rosser of the U.S. Army Math- 
hematics Research Center of the University of Wisconsin chaired the Com- 

25 mittee. The Report estimated that in 1964, colleges and universities 
";had about $250 million worth of computer equipment installed in those 
;'400 centers. Universities' annual EDP budgets were comparable to the 
ijcosts of running their libraries. (DX 5504, p. 15.) 

I! -437- 



z 

4* 
5 

! 

t\ 

3 

ia : . 

i 

12 | 
13 

U ! 
i 
t 

15 I 
16 

17 j 
IS 
19 
20 



2i ii 



edged, computers were becoming more and more important on the nation's 
campuses: 

"Campus computers are used by an increasing number of 
students either to do homework or laboratory problems, or to 
learn about the design and operation of computers themselves . 

» 

"Campus computers,- like laboratory equipment, are needed to 
do research* They increase the effectiveness of other scientific 
equipment and permit many scientific studies of a scope and depth 
heretofore unattainable," (DX 5504, p. 15.) 

While in 1957 computer costs represented only 3% of all university 

research and development costs, by 1963 the percentage had more than 

tripled to 10.04%. (DX 5504; p. 66.) 

But government funding was insufficient to support the 
growth in computing which universities were experiencing during that 
period. Computer equipment was expensive, and universities could not 
afford it without additional help. In 1963, for example, colleges and 
universities spent about $97 million on computers. About half of that 
came from federal sources, and colleges and universities themselves 
were able to pay for about 34%, a shortfall of 16% remained to be 
provided from other sources. (DX 5504, pp. 18, 21.) 

In order to make up that shortfall, colleges and universities 
turned to computer equipment manufacturers for help. (Morse, Tr. 
30965.)* The business equipment manufacturers had historically offered 



22 j * DX 5462, a listing of Requests for Computing Hardware compiled 
;|by the National Science Foundation, lists 366 proposals from 175 

23 ^educational institutions from 19 57 to 1967 asking computer manufacturers 
if or free or discounted eauipment. (DX 546 2, p. 20.) 

24 ;i " 



-438- 



L 
2 



special discounts to universities,* and that practice was continued. 
For example, when asked to explain why Burroughs offered educational 
discounts, Macdonald testified: 



"First of all, it appears that it's been an industry prac- 
tice for a very long time.. .... [A] long with that, the educational 

institutions appear to have grown accustomed to this practice 
and remind us of it should we forget, and it is practiced by 

5; our major competitors and it seems to sort of satisfy the general 
social pressure that educational institutions should be treated 

7 lj in a kind of special category as far as pricing is concerned." 
(Tr. 6986.) 



3 
9 

10 

II 

12! 



14* 
15 



Thus, the pleas were generally successful: 

"In the recent past, operating costs for computer centers have 
increased too rapidly for the usual university financing. . . . 
This difficulty has been partly alleviated by the generous educa- 
tional contributions offered by some of the manufacturers ..." 
(DX 5504, p. 20.) 



Helping universities acquire and use computers was clearly in 
the self-interest— or, as DeCarlo of IBM put it, "enlightened self- 
interest" (DX 7514, p. 8)**-- of computer manufacturers. The use of 
computers at universites was an important means of gaining the wide- 
ns ; spread acceptance of the new technology. It offered the promise of 
*7 'I overcoming some of the ignorance, fear and uncertainty about computers 

*;l 



13 ;j * For example, National Cash Register Co. offered educational 

i[ discounts on cash registers and accounting machines at least as early 

2CJ :| as 1929. (DX 347, p. 2.) Similarly, Raymond Macdonald, chief execu- 

'i tive officer of Burroughs, testified that the educational allowance 

21 I "practice was in effect when I joined the business in [the] mid- 

i 1930s." (Tr. 6986.) IBM offered educational discounts in the mid- 
22 '} 1930s on equipment to be used for teaching and research. (JX 28, II 
11.) 

j ** According to DeCarlo, "The evolving patterns of corporate support 
2- i of education predicate corporate giving on the basis of enlightened 

j self-interest, a concept that serves to illuminate the mutual nature 
25 l of corporation-education relationships . The long range interests of 
IBM and education coincide in important ways." (DX 7514, p. 8.) 

-439- 



by training the new generation in their use. 

There were more direct potential impacts .. The infant 

industry was suffering from an acute shortage of people who were 

» 

trained in computing,- educational, discounts would help- alleviate that 

shortage „ According to the Rosser Report: 

" [Educational discounts were] first instituted because the 
manufacturers realized that they would have trouble selling 
computers unless people capable of using them were available. To 
encourage the training of such people, manufacturers gave dis- 
counts to schools offering courses related to computers ; the more 
courses, the greater the discount." (DX 5504, p. 44.) 

Also, as more and more people became knowledgeable about computing, 

additional applications for computers would inevitably be created, and 

the market would grow. The Rosser Report described that phenomenon as 

follows : 

" [Ulniversities can draw upon the talents of their students, 
1 1| the best minds of each generation, at a time when these minds are 
alert, inquisitive, and full of fresh ideas. Because a university 
can bring these minds into contact with the computer in an atmos- 
phere conducive to research and imaginative thinking, it can 
.. I stimulate bold and original ideas for improving the computer and 
making better use of it. There is, therefore, great value in 
supporting such activity in universities." (DX 5504, pp. 28-30.) 



T \ 

3! 

2 



In addition, some people believed that computer manufacturers 
il would derive a positive public relations return from an active program 

ij 

j in support of higher education. DeCarlo of IBM believed, for example, 

i 
that "beyond fulfillment of 'corporate citizenship' responsibilities 



1 1 
•j there is significant potential for public relations return on the 

.(education support investment". (DX 7514, p. 6.)* 
3 ; 



i 



* Some people thought that such a "public relations return" would 
5 i include students who, having been trained on computer equipment of a 



;j -440- 

i 



1 

2 

a 

4- 
5 



For those reasons, IBM and other vendors offered a variety o: 
support to educational institutions. IBM offered educational allow- 
ances of varying percentages depending on whether the equipment was to 
be used for administrative or instructional purposes, and IBM also 
donated computer time to universities under circumstances that would. 
6 ; ensure that the time, would be made available to a wide variety of 
students** In addition, manufacturers, especially CDC, offered 
research grants** (Norris, Tr. 5647) , "buybacks" of computer time/ 



7 I 
S I 



10 

11 



particular vendor t would later be inclined to favor that equipment. 
ji (See, e.g., Hangen, Tr. 10448-49; Rooney, Tr. 11880 ("[W]e felt it was 
ji advantageous to have the students in the university become acquainted 
ii with computers by first utilizing RCA equipment.").) Other evidence, 

^2 ji however, suggests that any such advantage was more apparent than real. 
;! (See, e.g., Perlis, Tr. 2033; Morse, Tr. 30985; Andreini, Tr. 47880- 

£3 |l 82.) As Wright testified: 

; t 
^4 !j "If you train a person on the use of a computer, he has an 

easy time going to some other manufacturer's computing system and 
£5 II adapting to that particular computing system. The fact that he 
;j was trained on an IBM system does not lock him into an IBM system 
and he is, therefore, able to handle another system." (Tr. 12910.) 



16 
17 
13 
19 
20 
21 



* In the mid-1950s, IBM established data processing centers on the 
ji campuses of MIT and UCLA on the express condition that any student fron 
; any college in the Northeast could apply, for time at the MIT center, 
•and that similarly any student in the West could use the UCLA facility. 
:; (Hurd, Tr. 86421; see also Morse, Tr. 30965.) Almost 40 colleges and 
; , universities ultimately participated at the MIT center and over 60 
:! participated at UCLA. (DX 7514, p. 33.) 



** Norris of CDC defined a research grant as a situation in which 
"the educational institution pays the list price, but in a separate 
■transaction we sponsor a particular research program at our expense". 
22 ;; (Tr. 5647-48.) 



2,5 '.; / Norris said that a "buyback" meant that "Control Data had the 

^ right to use time on the machine for its own purpose in an amount equi- 

*4 valent to that expressed in terms of the monthlv rental". (Tr. 
,5988.) 

25 ; 

• -441- 

I! 
ii 

ii 



1 :j(Norris, Tr. 5988), and large cash contributions.* (Morse, Tr. 30980.) 
Larger discounts for academic or instructional use than for 

3 ! administrative use were not uncommon, and were consistent with the 

4 ildesire of manufacturers to encourage the training of people knowledge- 

i 

:i • 

5 !iable in computing. The Rosser Report observed that: 

I 

6 .1 "There are numerous cases where computer companies have given 

ij support in the form of generous discounts on the rental or pur- 

7 |l chase price; in many cases, this has been done on the condition 
;| that some computer time be made available for instruction. This 

3 si has ramifications similar to those arising from the NSF [National 
|l Science Foundation] support with its side condition." (DX 5504, 

9 j P-. 57.) 

1 1 
IQ IjLooking toward the future, the report argued t 

!! 

H Is "The computer manufacturers are gravely concerned with the ques- 
tion of education. For one thing, the shortage of programmers, 

^2 |j referred to earlier, has a dampening effect on sales of compu- 

ij ters . If the computer manufacturers could be assured that their 

^3 jl educational discounts would really support education, rather 

II than being used in good part as an indirect subsidy of government 

^4 !i research, they would be disposed to return to the more liberal 

j discounts of earlier days."** (DX 5504, p. 58.) 

15 ; 
15 i 
17 I 
13 ; 
19 ' 
20 
21 
22 



In restrospect, the educational allowance plainly accom- 
plished the goal of supporting the growth of the industry, as well as 



* In the early 1970s, for example, CDC donated $5 million to MIT, 
payable over five years. (Morse, Tr. 30980.) Morse, head of MIT's 
.Computation Center, was also a director of CDC but he testified that 
'there was no connection between that fact and CDC's contribution. ( Id. j 



** The reference is to the Carnegie decision (PX 1088) , a ruling by 
the Armed Services Board of Contract Appeals that universities had to 
pass educational discounts on to the government whenever computers 
acquired on such a discount were used in government-sponsored research. 
The Rosser Report criticized the implications and developing repercus- 

2j sions of the Carnegie decision and concluded: "In view of the pressing 
need for education in connection with computers at the present time, 

24 -this tendency of government officials should be reversed." (DX 5504, 
p. 45.) 



25 



-442- 



benefiting the society in general. In a draft report prepared for 
the National Science Foundation, Prof. W. F. Miller of Stanford Univer- 
sity concluded, that the educational allowance 

"was a very important form of support in the early years. It 
contributed immensely to the growth of the computing industry in 
the country* The computing industry grew in its most spectacular 
growth from the ground up. When the colleges and universities 
began to graduate engineers, scientists, business school 
graduates , etc . ,. who had been introduced to computing through 

7 if introductory courses and. often had taken advanced courses in 

computing, they began to introduce computer methods into their 

3 if respective businesses. This in turn stimulated the great demand 
for computers and the spectacular growth of the computer industry 
in the early and mid-1960s. There is no doubt that the colleges 
and universities who first introduced large teaching programs in 
computing would not have been able to support these educational 
courses on such an extensive scale without the benefit of the 
[educational allowance]." (DX 5500, p. 3.) 



10- 

II 
12 
13 
14- 
15 
IS 
17 



13 



Similarly, the President's Science Advisory Committee, 

writing in a 1966 Report titled "Computers in Higher Education", 

observed that: 

"Great good has been done through donated computers , obsolescent 
computers , huge educational discounts , grants for the purchase of 
computers and the struggles of enthusiastic men with inadequate 
machines." (DX 5476, p. 18.) 

Speaking of IBM's educational allowance program, Hurd tes ti- 



ts il 

-° it fied that educational institutions and society in general benefited 



from educational allowances : 



2$ ;j "First, because of that educational allowance policy they were 
able to afford the installation of general purpose computer 
systems, and having afforded it, they were able to support their 
instructional programs , support their research programs , and as I 
have indicated in my testimony, increasingly the use of general 
purpose computers supported research not only in the physical 
sciences but in the social sciences and in the humanities . So 
I in that sense the IBM educational allowance policy contributed to 

2- [ > society in general because of the research results and the 

:! instructional results." (Tr. 86 715.) 

i 

; ! -443- 



77 



25 : : 



3 



IBM's Educational Support Programs. IBM's support of educa- 
tion started with the beginnings of the company and it was originated 
and directed to a large degree by Thomas J. Watson, Sr. (DX 12150, 
pp. 17-18? DX 7514, p. 8.) 

Watson had a very strong belief that IBM should support 

educational institutions because " ' [t]here is no saturation point, in 

education."* (DX 12150, p. 18.) When IBM's SSEC became operational 

in 1948 (see pp. 24-25 above), Watson dedicated it to science and 

IBM allowed educational and research institutions to use the machine 

without charge. (Hurd, Tr. 86420.) At the dedication of the SSEC on 

January 27, 1948, Watson told the assembled guests: 

"It is with a mixed feeling of humility and confidence in 
the future . . . that I dedicate the IBM Selective Sequence 
Electronic Calculator to the use of science throughout the world." 
(DX 12150, p. 32.) 



Mr. Watson's strong belief regarding IBM's support of educa- 
tion and the mutual benefits which would accrue was the basis for 
IBM's continuing policies in this area. As Dr. Hurd testified, IBM 

7 ij "hoped to benefit from the expansion in the understanding, uses and 

■I 

5 ;{ users of computing . . . and I believed that all suppliers of general 
9 j purpose computers would benefit from its policies". (Tr. 86422.) 

G .1 He went on to comment: "they [educational institutions] were able to 

i 
1 | afford the installation of general purpose computer systems ... to 

support their instructional programs . . . research programs ... in 

i 

the physical sciences . . . social sciences and in the humanities". I 

I 

6 1 (Tr. 86715.) ! 
[5 j There are many examples of IBM's early support of educational ! 

! and academic endeavors. For example, in 19 24, Henry Wallace, who j 

i ! 

,1 -444- ! 



1 I later became Vice President, used IBM punched card equipment which had 

2 i been donated by IBM to Iowa State College, in order to do the research 

3 :} that led to his Ph.D. degree and later to the invention of hybrid 
ii 

4 i| corn. (Hurd, Tr* 86712.) During the 1930s, IBM supported Dr. Wallace 

5 i'JEckert's astronomical research at Columbia University and in 1945 
5 .j established the Watson Laboratory at Columbia. (Hurd, Tr. 86713; 

7 -iJDX 12150, pp. 11, 23.) In the 1940s, IBM supplied the machines that 

:l 
ft 

g.'ijWere used by the Carnegie Foundation for the Advancement of Teaching 

\\ 
q »J to evaluate test results from a nationwide survey. (Hurd, Tr. 86713.) 

it 
in !l In October 1955, IBM announced an educational allowance 

.- j; program for the 650 computer. An allowance of 60% off the rental 

.- :i price was available to educational institutions that offered courses 

,3 ;jin both scientific computing and data processing. (JX 28, fl 12.) 

The 60% discount provided a great benefit to universities. Perlis 

testified: 



14 
15 
15 
17 



"The 60 per cent discount that IBM made available to 
universities opened up digital computing in the universities 
in the sense that almost no university was able to afford 
or at least thought they could find funds of the kind 
required to establish a digital computer laboratory until 

IS • that discount became available, after which there were just 
a very large number of IBM computers, in particular IBM 

1$ : 650' s finding their way into the universities and forming 
the focus of university computer centers." (Tr. 2009.) 



20 
21 
22 



That educational allowance policy was "absolutely" one of the "principa 
forces which enabled universities to become competent in computing as 



soon as they did". ( Id. ) 

23 

In May 1960, IBM announced that a 20% allowance would be 

24 . 

roffered on all of its EDP machines, systems and features, leased or 
25 

"purchased, used for administrative purposes, and that a 60% allowance 

I: 

I: -445- 



would be available if that equipment was used for instructional pur- 
poses. (JX 28, 1f 13.)* 

IBM's allowance program remained relatively unchanged** 
.j|- until February 1963, when IBM, abandoned the administrative use /instruc- 
tional use distinction and reduced the 60% allowance to 20% on all new 
orders. (JX 28, fl 17.) 

When System/ 360 was announced in 1964, IBM left the percent- 
age of the educational allowance unchanged (at 20%) and made it 
available on System/360 equipment to colleges, universities and junior 
colleges. (JX 28, 11 19.) However, it soon became apparent that 
educational users who had second generation equipment installed under 
the much higher 60% allowance could not afford to take advantage of 
the price/performance improvement that System/360 offered because the 
educational allowance on System/360 was so much lower. (See, e.g., 
DX 12435, Armstrong, pp. 80-81.) 

In addition the competition for the business of educational 
institutions was especially severe during the middle 1960s, and most 
of IBM's competition offered high educational discounts and other 



I i 
I 



■j special arrangements such as lucrative research contracts. (See, e.g. 
iNorris, Tr. 5647.) Macdonald of Burroughs said that on occasion 

a i 



.j * IBM estimated that in 1962, it would grant allowances worth $24 
Z i million. (DX 7514, p. 30.) 

J 

j | ** Junior colleges and post-high school vocational institutions 

I became eligible to receive educational allowances for certain equip- 
,4. Iment in 1961 and 1962 respectively. (JX 28, «Mf 15, 15.) 



-446- 



L 
Z 

3 

4- 
f 

e 

7\ 

3 

9 

ia 

11; 
12 

14- ; 

15 

IS 



Burroughs' discount reached as high as 50% (Tr. 7534), and Norris of 

CDC testified t 

"[W]hen we first started the company, as I recall, educational 
discounts then were around 20 per cent. And then they increased, 
and at one period X recall their getting as high as 60 per cent.'* 
(Tr. 5648.) . 

In fact, CDC would often offer combinations of discounts, research 
contracts and buybacks in certain competitive situations. For example, 
at Battelle Institute, CDC offered a 20% educational discount on a 
CDC 6400, and in addition CDC gave Battelle $10,500 a month as a 
"buyback" of computer time. Similarly, at New York University, CDC 
offered a discount of $15,060 a month on a CDC 6600, and $14,400 as 
a "buyback". At the University of Illinois, CDC offered a 1604 
computer for 13 months on what CDC called a "100% Rent-free consign- 
ment". (DX 278, pp. 1, 6; see also Norris, Tr. 5989-91, 5993-95.) 

Plaintiff's witness Wright, a former Regional Director of 
Marketing for the GEM region in IBM's Data Processing Division, was 
asked whether IBM's educational allowance served any purpose from a 



^ I; marketing standpoint. He responded in part: 



-S || " [A] 11 vendors to some extent had educational allowances to my 

;j knowledge. All of the companies I have been associated with have 

-3 jj an educational allowance and this, in turn, permitted IBM to 

;| compete for business in educational institutions, in addition to 

2G ,| providing computering [sic] equipment at a lesser cost to those 

:i educational institutions." (Tr. 12958.) 

21 \ ■ 

\ As reported internally by IBM employees in 1964, Burroughs, CDC and 

22 j 

1 GE were offering a wide range of discounts, from 20% to 60%, along 

i with other significant considerations such as cash grants. (PX 
24. 1 

:! 2963, pp. 7, 9.) 

45 i 

: -447- 



During the period from 1965 to 1968, other computer manufac- 
turers were successfully marketing computers to universities . For 
example , DEC reported that its computers were used at Yale University , 
MIT, Stanford University, University of California at Berkeley, Oxford 
University, Harvard University and University of Wisconsin, among 
others. (DX 13846, p. 6,- DX 13847, p. 7.) SDS reported that users of 

ij its computers included Johns Hopkins University, Duke University, 
University of Delaware, Michigan State and UCLA. (DX 983, p. 10.) 
And Hewlett-Packard reported that colleges were using its equipment. 
(DX 11011, p. 10.) 

As a result, IBM and its competitors frequently became 
involved in highly competitive situations at universities. (See, 
e.g., PX 1824 (Berkeley); PX 1468 (University of Pennsylvania); 
PX 1558 (University of Colorado).) 

Even if IBM had discontinued its educational allowances 
altogether, it seems probable that other manufacturers would have 

;j continued them nevertheless,* and it is therefore hardly surprising that 

'{ to some extent IBM considered such allowances a competitive necessity. 

■j As a result, in 1965, IBM raised slightly the educational allowance on 

| System/360 computer equipment, and created a sliding scale of discounts 
f ': 
| ranging from 20% on the Model 30 CPU to 45% on the larger CPUs. 

\ 

i :l 

* When asked what Burroughs would do if IBM were forced to discon- 
; j tinue its allowances, Macdonald said, "I believe we would consider it 

j carefully, and were that to happen today I think we would probably 
1 ) continue the practice." (Tr. 6987.) Moreover, Macdonald said that 

| "I suspect that for the remainder of the industry that the practice 
j : | would continue . " ( Id. ) 

j -448- 



L 
Z 
Z 

5 



Even though that increase helped colleges and universities 
to acquire the newer System/360 computers, and even though it enabled 
IBM to be. more competitive, a great debate ensued, within IBM as to 
whether high educational discounts were the most appropriate way for 
IBM to support education.* Some favored continuing the discounts; 



5 ;! some favored raising them; others favored lower ding or even eliminate 



7 
3 
9- 

a 

12. 

!3 '• 
14 



ing them.. Still others favored massive efforts in support of educa- 
tion. 

As examples of the differing opinions, T» V. Learson was 
quoted as saying i 

"We [IBM] have two objectives in this [E.A.] program: the 
first to get university customers back up to paying full 
rentals or as close to it as possible in the long-run; the 
second, to get more revenue in the short-run, i.e., 1966." 
(PX 1652, p. 1.) 

Herman Goldstine, when he was Director of Scientific Development at 

DPD Headquarters, observed that "the educational allowance was ori- 



{[ ginally introduced by IBM as a matter of enlightened self-interest 



and the expressed" intention was to further the training of young 
people in the use of the computer." He went on to recommend "that we 
should substitute for the EA or for most of it a cash .grants program 



17 \ 
IS ■ 

is ;, 

: or a value-received program. Perhaps we will always want a 5% discount 

20 ':! 
1 

I for psychological reasons." (PX 1679, pp. 1, 2.) Armstrong took the 

21 \ 

;] position in November 1965 that the educational allowance program be 

22. 'i ■ • 



\ * That debate was fueled by the Carnegie decision, referred to above, 
24. ! which had the effect of passing the manufacturer's discounts on to the 

(government, a result which manufacturers had not intended. 

25 .1 

\ 

!l -449- 



1 left alone, i.e., unchanged from the March 1965 position. (PX 3871: 

2 Armstrong, pp. 145-46; see also Wright, Tr. 12912-13; compare 

3 PX 1661, p. 1.) 

4 The result of the debate was a corporate decision gradually 

5 to reduce the allowance to 10%. (PX 1706, p. 2; PX 1745, p. 2? 

6 px 1746, p. 2.) In 1966, the educational allowances on most 

7 equipment were reduced by about 10% of the price of the equipment. 

8 (JX 28, 11 26) as the first phase of the larger reduction. Xn 1969, 

9 the allowance was reduced to 10% as planned (JX 28, II 29), where it 
10 remains today on most products. 
11 
12 
13 
14 
15 I 



16 S 

17 

18 

19 

20 

21 

22 

23 |j 

i; 

24 !| 

II 

i: 

25 i, 



-450- 



L; 39. IBM's Unbundling. 

2 l a. Introduction. Before discussing IBM's unbundling in 

3 ;j 1969, it is useful to review briefly the causes and effects of bun- 
4.|dlingin the 1950s and 1960s. Bundling, as used in the EDP industry and 

3 it with respect to IBM, is "the offering of a number of elements that are 

i' 
3j considered to be interrelated and necessary from a customer's point of 

;( 
j i( view, m the computer field, under a single pricing plan, without 

detailing the pricing of the component elements themselves." (R. 

Bloch, Tr. 7603-04.) The elements which were offered without a 

separate price were non-hardware elements such as education, software, 

i 

systems design, and maintenance.* (see above, pp. 56-67.) As de- 
scribed above (pp- 53, 56-67), the provision of such support services by 
manufacturers greatly facilitated the marketing of their equipment to 
users by reducing the users' risks in installing that new, unfamiliar, 
and expensive object, the computer. (See R. Bloch, Tr. 7751-54; 



3 

9 ., 

i 

10 i 

u ! 

12 1 

T1 



14; 
I 



I Norris, Tr. 6058-59; McCollister, Tr. 11041-43; Welke, Tr. 17380-81, 

la " 

17 



i 



* Maintenance was included in the lease prices for equipment 
that the manufacturer continued to own and the user leased. 
19 ;| Maintenance was priced separately for purchase customers. (See 

;| Spangle, Tr. 5094-97 (Honeywell); Macdonald, Tr. 6980 (Burroughs); 
20] Weil, Tr. 7087-88, 7099-100; R. Bloch, Tr. 7804 (GE) ; McCollister, Tr 
i 11461, 11476-77 (RCA).) Manufacturers have strong incentives to 

21 1 provide such maintenance in order to protect their property. (See 

j Norris, Tr. 6069-70, McCollister, Tr. 11476-77; Vaughan, Tr. 21732-35 

22 I The significance to the users in the early days of maintenance 

1 being included to lease customers, however, was similar to that 

23 i of the bundling of other support services — it increased users' 
[willingness to experiment with this new equipment and helped to 

24 1 assuage their fears and minimize their risks. (See Welke, Tr. 

"19225-28; see pp. 53-67 above.) 



?« 



-451- 



17343-46.)* 

As a consequence, virtually " [a] 11 the computer manufac- 
turers marketed on a bundled basis" during the 1950s from the Univac 
I on. (Goetz, Tr. 17500-01; Spangle, Tr. 5092; R. Bloch, Tr. 7604; 
McCollister, Tr. 11042-43; see also Norris, Tr. 6066.) 

At IBM, the provision of bundled support began before the 
installation or even the acquisition of a computer by the customer. 
Such support was viewed both inside and outside IBM as an essential 
part of the marketing effort. The IBM systems engineer (SE) was 
"part of the marketing team" (Akers, Tr. 96554-56) and would 
j| assist in the preparation of the proposal made to the customer. 

( Id . ; Enfield,** Tr. 19908.) The IBM salesmen drew on them for tech- 
!j nical support. It was the systems engineers who "had the implied 
; | responsibility of . . . developing systems to make sure that the 



| \ * In formulating the provisions of the 19 56 Consent Decree 
:j concerning IBM's obligations to customers that purchased equipment 

r ! | from IBM, the Department of Justice apparently recognized the benefit 
•1 to users of the support provided by IBM without separate charge. The 

^ .! January 25, 1956 Final Judgment obligated IBM "to offer to render, 
;| without separate charge, to purchasers from it of tabulating or 

• '! electronic data processing machines the same type of services, other 
"j than maintenance and repair services , which it renders without 

i ■' separate charge to lessees of the same types of machines". ( U.S. v 
Vj IBM, [1956] CCH Trade Cases § 68,245, Part VI, §(a), (S.D.N.Y. 1956) .: 

r ! 

• I ** At the time of his testimony in 1976, Enfield was President 
? * of The Computer Software Company. (Tr. 19841.) Between 1964 and 

• j 1969 Enfield was employed by IBM, first as a systems engineer and 
. j then as a Product Administrator in Data Processing Division head- 
3 i auarters. (Tr. 19843-44.) 



-452- 



machine was put to good use". (Welke,* Tr. 1700 9.) They worked 
with customers to define requirements and in system design, develop- 
ing approaches to problems, also engaging in customer education and 
training and in programming. Such work would sometimes continue 
after installation. (Welke, Tr. 17007-10, 17069-70, 17372-73.) 
In short, the systems engineers were responsible for "making sure 
that the customer was indeed implementing the targeted applications, 
the 'business applications, and doing the job properly and being of what- 
ever assistance we could to make sure that the machine was . . . 
performing properly". (Welke, Tr. 17010? Akers, Tr. 96555-56.) 

Other firms in the industry also provided those types of 
services as part of their marketing efforts. McCollister testified 
that it was "normal for some fraction of the time of the [RCA] market- 



1S 
17 



IS;: 



■*■- j * At the time of his testimony in 1976, Lawrence Welke was President 
,j of International Computer Programs, a firm providing "an information 

2G 4 service to the computer software product marketplace" by publishing 
i catalogs of software products and by conducting seminars on buying 

— •jand selling software. (Welke, Tr. 17003-04.) Welke's first job in 
•j 1954 was with General Electric and he had the responsibility of 

22 j installing a punch-card system in GE's production department. Between 
j 1956 and 1963 Welke worked at IBM as a systems engineer for three 

23 j years and as a salesman in the Data Processing Division for four 
jyears. Between 1963 and 1968 Welke was with a consulting firm and a 

2- ) bank as head of their automated customer services division. (Welke, 

JTr. 17004-07.) 
25 j 

i -453- 






"salespeople, maintenance people and systems analysts and programmers, 
technical people" as "a normal and as a necessary part of the success- 
ful sale and installation of computer equipment." (Tr. 11370-72.) 
I' In 1972 Ray Macdonald, President of Burroughs, stated that: 

"A major element of the marketing effort in our industry 
is support activities. It is important to note here that our 
industry's involvement with its products lasts throughout the 
lives of those products. It starts with semi-finished raw 
materials, continues through intermediate and final manufac- 
turing processes, and extends to a full range of services in 
support of the product throughout its use. 



"At Burroughs, we developed a worldwide capability in 
excellent technical support of our products very early in 
the traditional product period. With the introduction of 
the computer, we have significantly extended our support 
operations by adding the new dimension of supporting the 
I customer in his use of the product. This includes systems 
y j planning and installation support, and perhaps most important 
of all, the support of the customer in his application soft- 
ware requirements." (DX 426, p. 12; see DX 427, p. 4.) 



I 

c 

a 

7 | 

Si, 

»1 

: account varied and not in any simple way. As Akers testified, 

:j systems engineers at IBM "were a scarce resource within the branch 

4 office" so there was an attempt "to manage the technical talent in 

J a way that was most beneficial in [IBM's] sales efforts and installa- 

2 i 

| tion efforts with our customers." Systems engineers "were allocated 

J- "i 

ion the basis of how much assistance a particular customer needed at a 



Similarly, in 1961 NCR reported to its stockholders that its "marketing 
organization . . . provides necessary programming aids, training 
courses for the customer's employees, technical assistance on site 
preparation, and other supporting services of various kinds." (DX 
402, p. 10.) 

Obviously, the amount of SE services needed at a particular 



particular time; the degree of experience that the customer had; 



; -454- 

i 
i 



L 
2 

3 

4- 

3 

y 
3 

9- i 

ia 

xi 

iz 

13 

. /* 

17 
18 



IS '■[ 



* i 



i 



whether or not that customer perhaps required additional educational 
effort because he or she was installing a new computer system or 
computer system for the first time. It was an effort to try to use 
that resource as productively as possible in pursuing the quota 
objectives that the branch office had." (Akers, Tr. 96555-57; see 
Enfield, Tr. 19878-79, 19886-88; DX 4793.) Systems engineers were 
assigned to customers on the basis of "who needed the work done and 
what had to be done to make it a successful installation". (Welke, 
Tr. 17017.) 

Systems engineering services were provided to familiarize 
users with computers and to ensure that the user, if he chose to 
acquire a computer, used it properly to solve his problems. Such 
service relieved users from some of the risk of acquiring a computer 
in order to induce them to acquire it in the first place. But, in 
relieving customers of such risks, IBM, like other manufacturers, 
assumed them. By giving the users "a predictable cost that they 
could budget against" (Welke, Tr. 19225-26) , the manufacturer took 
over the uncertainty in cost resulting from unforeseen variation in 
user needs. 

Concomitantly, manufacturers stood to gain (by lower costs) 
( if over time the customer required less or no assistance. In the 
J long run, the reduction in customer needs would be accomplished in 
part, as it turned out, by the provision of increasingly sophisticated 
operating systems relieving customer programmers of a number of 



24. 1 

j complex tasks, but it could also be accomplished in the short run by 



-455- 



* 

T 

\ 
} 

} 

L 

? !' 
'I 

2 



training customer personnel in the tasks which software had not yet 
taken over and in the use of the software- hardware combinations. 
Thus, according to the IBM "guidelines" concerning programming, 
"Systems Engineering personnel were to clearly encourage self-suffi- 
ciency among the customer [sic] in his programming capabilities with 
regard to application programs." (Enfield, Tr. 19862; see also 
Welke, Tr. 17373-75.) Such self-sufficiency "was a self-serving 
objective. The objective to enable the user to provide more of his 
own support would enable an SE to perform less of those functions", 
freeing the systems engineer for other assignments. It would also 
make the customer more efficient if he did not have to depend on 
others. (Enfield, Tr* 20249-50.) 

In November 1962 Prank T. Cary, at the time Vice President of 
Field Operations for the Data Processing Division, put out guidelines to 
IBM executives, regional and local management and sales representa- 
tives and systems engineers saying that it was IBM's "responsibility" 
to provide to its ■ customers "the assistance they need to install and 
obtain the results from the use of our equipment that we have outlined 

'{in our proposals to them". Among IBM's responsibilities were the 
9 I 

'\ "[e]ducation of customer personnel" and the provision of "[tlechnical 

5 i 

j guidance" in "the use" of IBM equipment and in "programming and test- 
II ! 

jing". Similarly, in order to underscore to the recipients the extent i 

2. ! ! 

! to which IBM was committed to having the customer assume responsibility, • 

'•* I i 

'the guidelines emphasized that it was the "customer's responsibility" ! 

14 : I 

| to "[w]rite his own operating programs", " [w] ire the necessary control 
?« i ! 

I ; 

! i 

\ -456- 



I 

2 
3 

4-i 
f 

7 
3 



panels", " [o]perate the equipment" and "[p]rovide for the physical 
installation of the equipment". (DX 4793.) It is worth noting 
that these guidelines applied only to lease customers and to "the 
first user of purchased equipment" to whom IBM felt it to be its 
"responsibility to provide . . . the assistance they need to install 
and obtain the results from the use of our equipment that we have 
outlined in our proposals to them". ( Id. ) IBM, quite naturally, was 
making this marketing support available to its customers and not to 



5 ; users that acquired IBM equipment from other sources. 



10 

12 

13 



Ralph A* Pfeiffer, Jr.,* described IBM's philosophy as follows 

"What we were trying to do was to insure customer's 
profitable use of the equipment. The Manager has a certain 
stable of talents; he had a customer set that he had to 
support and he tried to make the most productive, efficient 
use of that cadre of personnel. 



"We are trying to supply a service to a customer. We 
are trying to have that customer make profitable use of his 
equipment. And if he is unable for some lack of whatever it 
might be, education in a certain area or a certain person who 
9/m . ' he relied on left and he was caught short, we try to supply 

10 ij that missing ingredient until he is able to handle it himself. 
■ We tried to train him. 



"We certainly were interested in having him be capable 
~-S |j of running his own installation in a profitable way. Whatever 

:j that required in the way of training somebody or supplying that 
1$ :{ piece of education that was missing, I hope I operated accord- 
ingly." (Tr. 16019-20.) 



i 
21 \ 



4 

22 i 

t 
.- i 



The policy of building self-sufficiency in customers, however, 



j *Mr. Pfeiffer is an IBM Senior Vice President and Chairman of the 
•; Board of IBM World Trade Americas/Far East Corporation. (DX 30 74, pp. 
2- 142-43.) At the time of his testimony he was an IBM Vice President and 
I the President of the Data Processing Division. (Tr. 2963-64.) 

^5 ! 



-457- 



carried with it an end to the practice of not charging separately for 

such services. When enough customers became self-sufficient and when 

changes in hardware and software ceased to require them to be taught 

i.J very new ways of operating/ i't would no longer make sense to bundle. 

By increasing self-sufficiency in customers, IBM created a growing 

group of customers who did not require the bundle. The exact date 

r j| on which that group was sufficiently large that it made sense to un- 

bundle and provide the formerly bundled services at separate charges 

for those who wanted them is a matter of judgment. As we shall see, 

in IBM's judgment it came in 1969. 

b. The Continued Demand For Bundling in the 1960s . During 

the 1960s, for the same reasons as in the 1950s, most users continued 

to prefer the bundled offering. (We Ike, Tr. 19230; R. Bloch, Tr. 

7751-54.) The demand for such support was not restricted to new 

customers unfamiliar with computers. Even in selling to "the large, 

I established user", such services would be required "to some degree". 

;i "[T]here would always be areas which are unfamiliar to even a relatively 

j sophisticated customer. The fact that he was graduating from some 

i smaller system to say, a larger . . . system which might involve 
i 

| communications, this communications area would be the first time for 
,i 

j that large customer. ... So even with sophisticated customers these 
1 •• 

j kinds of support were required." (Beard, Tr. 9944-46.) 



> 

1 
I 

1 i 

I 

L 

e? 
■* 

s 

7 
S 
S 
.0 



_£. » 



] The demand for support services continued in the 1960s as 

users were rapidly exploring new computer uses and as software 
improvements and architecture changes were occurring at a rapid rate. 



La 



M 



-458- 



Li 
2 ; 
1 1 

4*: 

5 

e 

7 
3 

? 

10 
II 
12 

13 
14 



iO < 



As Beard testified: 

"... most of the customers we [RCA] were dealing with 
in the time frame of 1960 to 1970 -were not thoroughly experi- 
enced in the use of data processing equipment. The field had 
gone through a very dynamic growth. It faced new technology, 
a new set of programs imposed upon the business organizations 
that used computers. So a lot of people felt they were on 
very shaky grounds. They were not sure of themselves." 
(Tr. 8497.) 

New products and new ways of doing things were being introduced requir- 
ing customer training, programming and systems design services, 
imposing additional demands on the manufacturers. (McCollister, 
Tr. 9647-53; DX 69, pp. 3, 5? see also DX 98, p. 12.) As Withington 
observed in 1968: 

"Programmers and system analysts are in inadequate 
supply. This shortage has existed for years and shipments 
have nevertheless grown, but in one major respect the problem 
is worsening. The advanced, integrated applications many 
users wish to implement are novel and very complex and require 
much more creative, high-level system analysis than the 
simpler, second generation applications did. Since experi- 
enced system analysts are in the shortest supply of all, this 
pressure may have an increasing effect. 



"The increasing complexity of the third generation 

'] hardware and' software (a necessary corollary to its increased 

i ' |l capability) makes it difficult for the average user to under- 

I stand and use. It may take longer than it used to for users 

^ i to fully exploit the equipment they are currently installing: 

;j many users will not be able to use anything larger or more 

•^ j complex for a number of years." (PX 4833, p. 9.) 



20 j Thus, although users would eventually, become familiar with 
i 

21 | the architecture of System/360, the sharp increase in complexity 

j. 

22 "\ as users moved from second generation equipment to System/ 3 60 tended 

23 j to offset the gains from previous experience. Users were being 

2- 'i trained and retrained to use more complex equipment in increasingly 

25 j sophisticated ways and the bundled IBM offerings were all the more 

t 
■i 

.j -459- 



important to the System/360 user* (See Welke, Tr. 19617-18.) For 
example, Welke testified that System/360 "represented a new level of 
hardware technology . . * it represented a new level of software 

technology with its systems software environment and the very way 

» 

that you approach programming and processing. It caused a complete 

change in how people approached the task of data processing."* 

( Id. ) In January 1964, F. P. Brooks, Jr., wrote that: 

"The breadth of System/360 and the number of innovations, 
particularly in gross systems concept, will require 
substantial lead time between announcement and proper 
installation. . . . The sheer amount of new abilities, new 
options, new specifications, and new prices will require 
time for the customer to assimilate. A major education 
program for IBM field personnel and customers must intervene 
between announcement and successful installation." (DX 1172, 
pp. 1-2; see also Withington, Tr. 56591-93; DX 4815.) 

As we have seen in the discussion of System/360, all of 

this happened — and more — with the result that IBM was compelled to 

expend tremendous effort and expense to install and support System/360. 

As other third generation equipment began to appear, other manufacturers 

found requirements for support services growing as well. SDS told its_ 

stockholders in its 1965 Annual Report that 



r 
i 

U 

2 
3 

7 

3 •! 
; | "Ct]he character of the computer market changed substantially 

a j last year as the result of advances in both the understanding 

•j of the technology and in the manner in which computers should 

a ! 

i 

T i 



2 ;| * "By [1968] the marketplace had acclimated to these new hardware 
1 technologies and software technologies. By the same token, the 

3 j product, the 360 system, particularly with reference to the software 
j involved, the system software, had settled down and achieved a 

m 'j respectable semblance of predictability." (Welke, Tr. 19617-18.) 



-* i 



-460- 



L 
2 

Z 
4- 

S 

7 i 
3 

9 

10 

II 
12. 
13 

14- 

15 
17 
IS 
IS 



be employed. . . . During the past year increasing emphasis 
has been placed by management on providing complete service 
to SDS customers both before and after installation. To 
this end, technical staffs and applications programming, systems 
engineering, customer training and maintenance have more than 
doubled in size and in the scope of their activities*" (DX 981, 
pp. 4-5.) 

Such increases continued for the next few years. (DX 982, p. 3; DX 983, 

pp. 16-17.) 

Similarly, RCA found that the introduction and installation 

of its new Spectra series created large user demands for assistance. 

(McCollister, Tr. 9649-53, 11403-06.) NCR told its stockholders in 

early 1966: 

"As the trend toward fully integrated business systems gains 
momentum, NCR's opportunities for growth and greater profit- 
ability can be expected to increase proportionately. Full 
realization of these opportunities will require an aggressive 
continuation of the program of recent years. To this end, 
additional expenditures will be required not only for further 
product development efforts but also for training sales and 
service personnel and for providing the many supporting 
services essential to the successful marketing of advanced 
business systems." (DX 368, p. 3.) 

It was a view which NCR was to reiterate as time went on. The 
following year, it stated: 



"Today, a . . . requirement for future success in the 
marketplace has arisen; that is the need for business equipment 
suppliers to provide additional guidance to customers in the 
utilization of new technologies for operating their businesses 

20 | more profitably. For in the final analysis, the effectiveness 
■j of today's sophisticated information systems depends upon a 

21 ;J - full understanding of their potential at all levels of manage- 
j ment. To this end, NCR's educational programs are being 

22 1 designed not only to prepare sales representatives to install 
] advanced systems, but also to provide counsel and training in 

22 ; management sciences." (DX 370, p. 5; see also p. 19.) 

i 
24. Sin its 1967 report, just after the announcement on March 2, 1968, 

25 I of its new Century series, NCR reiterated this position, 
i 

! 

:! -461- 



3 

1 

T 

z 

m 

3 

4» 



"In addition to offering outstanding equipment, meeting 
the demands of the market today also requires: 

"1. Expert diagnosis of customers' current and future 
business information requirements , based on broad systems 
knowledge and experience. 

"2. A complete range of supporting software, including 
standard programs for many applications and in-depth training 
of customer personnel. 

"3. Continuing support of every installation, with 
upgrading of both system and equipment as customer requirements 
change . 

"The company's marketing strategy is based on providing 
this full spectrum of customer services." (DX 366, p. 3.) 

Similarly, one of Sperry Rand's major objectives in 1962 and 1963 

was to "give increasing emphasis to our computer service and 

marketing". (DX 69, pp. 3, 5.) It was an "emphasis" which R. E. 

McDonald was to look back on in 1973 as "[o]ne of the main factors" 

behind Univac's success. (DX 98, p. 12; DX 65, p. 2.) 

c. IBM's Unbundling Announcement. On December 6, 1968, IBM 

;| announced that it expected "to make changes in the way it charges 

;) for and supports its data processing equipment" during the following 

7 il 

;| year. (PX 3390.) It announced its decision in detail on June 23, 

8 3 

: 1969, with the changes effective immediately for new orders and 

I effective January 1, 1970, for customers with machines installed or 

I on order. (PX 3351; PX 3352.) Basically, the announcement instituted 

- ! 

•ij charges for systems engineering services and education and for new 
•2 ) 

■ "program products, as distinct from system control programming". (PX 

3351, p. 4.) Programs then available from IBM's library continued 
l± \ 

.{to be available as in the past at no separate charge. I3M also 

IS ' ] 

] 

\ -462- 



L 
2 

3 

4- 

5 
7 

a 

9 

10 
11 ] 
12 
13 

14* ; 

15 

16 



offered to engage in contracts assuming "responsibility for the 
performance of specified tasks in the areas of systems design and 
analysis, application and program development and systems install- 
ation and evaluation" . ( Id- ) No change was made in the way in which 
maintenance was provided, maintenance on purchased equipment continuing 
to be available at a fee and maintenance on IBM-owned equipment 
leased to users available without a separately stated charge. (PX 
3351; PX 3352.) IBM also reduced its prices by 3%, stating that 
this reflected its "best approximation" of the expenses which would 
"no longer be provided for in prices of currently announced equipment". 
(PX 3351, pp. 1-2.) 

There were a number of reasons for the announcement. 

First, IBM, like others, was feeling the strain of standing ready to 

supply services on demand without an extra charge in an increasingly 

complex environment. IBM "stated that — as a result of fast-changing 

i data processing market conditions — the need for increasingly complex 



and comprehensive systems support is growing more rapidly than antici- " 
I pated. In addition, new support requirements are arising from leasing 
;j companies and other owners of IBM equipment as they relocate and 
^reapply their systems." Such demands for "new and additional forms of 

i support services" were expected to continue to grow. (PX 3390, pp. 1-2; 

21 ! . 

IPX 3351, p. 3.) As would be expected in a company accepting the risks 

I and burdens bundling entails, there was a recognition within IBM of 

22 j 

[the increasing costs of providing software and support. During the 
24 i 



25 



j early and mid 1960s persons within IBM observed that programming 



-463- 



expenditures were "skyrocketing" and "increasing dramatically" (PX 
2804A, pp. 1, 2; PX 2805A, p. 1; PX 4053, p. 1), and attempts were 
made in 1966 to quantify the return to IBM on programming expend!- 
' tures. (PX 1748, p. 11.) Cary testified that increasing demands of 
customers for education led to separate pricing of certain education 
offerings and that IBM was "always looking for ways of reducing the 
cost of systems engineering". (JX 57, p. 2.) 

The general problem of cost escalation was magnified by 
the special problems associated with installation of System/360. As 
we have noted earlier, because of the unprecedented — and unantici- 
pated — success of System/360, IBM had added new people to its marketing 
division. (See p. 372 above.) The training of such people, the 

support required by users to effect their conversion to the new and 
sophisticated operating system software associated with System/360, and 
the problems which IBM encountered with some of the 360 software caused 
IBM to devote an enormous portion of its resources to supporting the 
j installation of System/360 and making sure that customers were able 

it to do their work during the transition phase. (See pp. 369-72 above.) 

i 

j The result of this, however, was that levels of support far greater 

5 :| 

| than ever before required were demanded of IBM. The cost of providing 

m I 

i such support had to be borne directly by IBM itself, but in the long 

■J run, of course, it would have to be absorbed by IBM's data processing 
2 ; i 

| users . 

2 1 

! At the same time, by 1969, in part as a result of I3M's 

4 1 

i policy of encouraging self-sufficiency, there had developed a group of 
el" 



-464- 



| 

z< 

si 

i 

5:1 
I 

7; 

3 I 

9 I 

i 

m 

n 

12 

a 

15 

* *•- 

17 
IS 
13 



relatively efficient and sophisticated users who would accept much 

more of the risks of computers and Were willing to do much of the 

support in-house* As Welke testified: 

""The more sophisticated users, and the ones who had the 
best-run or the best-managed shops, for the most part were 
ready to accept the idea of unbundling, because I think they 
saw in it a chance to be more cost effective in their entire 
data processing operation." (Tr. 17172-73.) 

T. J. Watson, Jr., testified: 

"We had some very sophisticated customers by this time, 
Lockheed, Boeing and others, who felt that they were better at 
performing, some of these services than we were. They felt it 
onerous to pay for them when they, themselves, could do it in 
their opinion better." (Tr. 16602.) 

Another reason for IBM's announcing unbundling in 1969 was 

that, by that time, the notion of charging for software and services 

had become relatively accepted because of the entry and success of 

software houses. (See below, pp. 851-65. ) That had not always been 

true, however. From the early days of the computer industry up until 

the late 1960s software was generally looked upon as something other 

than property that could be appropriately charged for. "For the 

longest time, computer programs were looked upon as an intellectual 

product, but not necessarily having proprietary value." (Welke, Tr. 



: ( 17361-62; see also DX 1096, pp. 1-2.) This led many people to believe 

20 : 

i that in fact most users were not willing to accept the notion of 

21 :| . 

i software as a "product" in the 1960s. (See Welke, Tr. 17093-95, 19180- 

J 82.) This view was both illustrated and reinforced by the free 

23 =| 

interchange of software that was characteristic of this period. (See 

24 "■» 

i below, pp. 856-58.) But, during the 1960s software houses began to 

25 i 

I charge for software products that competed with IBM's unpriced offerings 



-465- 



and by 1969 " [t]he industry had developed to a point where many of those 
services were available, separately, and outside." (Watson, Tr. 
16601; PX 3351, p.. 3; see pp. 858-59- below. ) .* .As a result, IBM began to 
if believe that, for the first time., there might be business opportunities 
in selling software and services separately. (E.g., PX 3351 r p. 3.) 
Under such circumstances, it was possible for IBM to stop offering 
such services, which it was finding "onerous", under the bundled system. 
IBM's Chairman testified that it "seemed like an appropriate time, 
from a business standpoint of view, to open the matter up in the way 
that we did". (Watson, Tr» 16602.) 

Not surprisingly, customer reaction to IBM's unbundling 
r I announcement varied. Some relatively more sophisticated customers 

welcomed unbundling; others, generally the relatively unsophisticated, 
were less happy. Welke testified: 



:l 



t ' 

3 

r I 
i 



"The initial reaction was — it varied. Some people were 
very happy with it. A good number of them that I came in con- 
tact with were anything from hostile to total disbelief as well 



"I think it depended on the sophistication of the user. 
The more sophisticated users, and the ones who had the best- 
;j run or the best-managed shops, for the most part were ready to 
3 ;l accept the idea of unbundling, because I think they saw in it 
|j a chance to be more cost effective in their entire data process- 
' : ; ing operation. 






T 

j 

, 

.| * One of the reasons for this was that industry practices had emerged 
3 •} which gave sellers some assurance of protection of proprietary program- j 

j ming from plagiarism. (Welke, Tr. 19211-13.) Welke testified that j 
A I "many sellers" of software at one point, and "to a very limited degree" j 

;! still, feared that their software would be plagiarized because of the j 

S j ease with which programming can be copied and the inadequate protection ! 
i 

t ; 

I 

i ; 

:! -466- 



z 

3 
4- 

3 

«• 

a 

7 

3 

? 
10. 
II 
VL 
13 
14 
15 

^ , 
17 \ 
IS : 

aa :! 



"The ones who didn't want the unbundling or who were 
against the idea I think in some cases were also the ones that 
were getting, . . . more than their normal share of IBM's 
support and systems engineering and programming as well . " 
(Tr. 17172-73.) 



. Similarly: 

" [Ulsers, even in 19-69, when they heard about unbundling, 
were reluctant to accept it or were hesitant and in some cases 
even hostile to the idea. At that point in time users were 
beginning to get a pretty good idea of what some of their cost 
elements were and the more sophisticated, more advanced 
users had a way of breaking out cost, elements in thei'- total 
computer operation, identifying them, and controlling them. 

"But for a lot of users, there were still many, many 
unknowns in their data processing operation, things that they 
didn't know could happen, they had no way of anticipating, and 
I think they wanted the assurance that bundling, in effect, 
offered them, that one way or another, if and when the unknown 
occurred, they'd be covered. It was an insurance policy in 
many respects." (Welke, Tr. 19226.) 

As might be expected, reactions of other manufacturers varied 
also. Bundling had been a practice desired by users. Users' needs 
changed over time as they became more sophisticated and self-suffi- 
cient but this was a continuous rather than a discrete process , and 
opinions, even in. 1969, could very well differ as to whether the time 
had come to make the changeover. For many companies the decision 
;j whether or not to unbundle was not entirely a foregone conclusion. 



22, j afforded software through patenting and copyrighting. (Tr. 19211-13 
!j This was recognized within IBM during the mid-1960s. In 1965 it was 

22. | stated within IBM that "an overriding factor against unbundling 
: j [certain programming] is our present inability to protect the pro- 

23 1 prietary use of our programming systems". (PX 1651, p. 6.) In 1965 
I R. H. Bullrsn, then an IBM Vice President and Group Executive, wrote 

24 i that: "We must settle on whether or not, and to what degree t 
\ we can protect programs before we can deal adequately with 

25 I the question of selling them". (DX 1031.) 

I 

! 

t -467- 



1 

2 
3 

r 

6 
7 
8 
9 

10 

ii 

12 

i 

13 
14 
15 
16 
17 
18 
19 
20 
21 
22 



Spangle, of Honeywell testified that Honeywell did not 
follow IBM's lead for a number of reasons. It was not set up adminis- 
tratively to charge for the separate items and enforce their collection 
throughout the field; it was not certain of the contractual arrange- 
ments it had. with its existing customers; finally, Honeywell "hoped to. 
gain some temporary market advantage . . . because we thought there 
would be quite a bit of resistance to this change by the customers and 
prospects, and that because of that we might be able to get some 
customers that we otherwise would not have been able to get". (Tr. 
5086-87; see also Withington, Tr. 56786-87.) 

Univac had similar reasons for not unbundling when IBM did. 

McDonald testified: 

"Actually, we felt that there would be considerable 
anxiety in the marketplace as the result of IBM's decision and 
announcement to unbundle, and we felt it would be to our 
competitive advantage to maintain our previous pricing policy 
so that we could go to the customers, potential customers of 
IBM, and say to them that we would offer you these services 
which we have in the past under the same pricing policy, and 
you know what you will be getting from us, and under the IBM 
unbundled pricing policy, only time will tell what your real 
prices will be; and I think this was effective, at least for a 
period of time* 



" [W] e did see some increase in bookings over what we 
expected our bookings would have been had IBM not changed their 
policy . . . which we attributed to IBM's unbundling." (Tr. 
2896-97.) 



Similarly, McCollister of RCA "recommended that RCA 
I; should continue in the business by continuing to offer bundled 
24 || services". He felt this to be "to the benefit of RCA in its 

relationship with its users .... [G] iving assistance to the user 
as required could lead to and usually did to a more effective use 

-468- 



25 : 



Li 
2 



3- 



of that equipment by the user and gave RCA, therefore, a stronger 
installation and to the extent that the equipment was on rent, 
insured more completely a continuation of the rental income." It 



4»l r brought in more money than unbundling. Further, "I felt, this had been 



a sound business policy for the IBM. company for a long time and just 
because IBM . . . decided that they would change, I did not see at 



7 1. that time that this was a reason for the RCA company to change and do 



3 

2 : 

ia 

11 : 

12 

13 

1^: 

15 

IS ! 

17 

l*| 

2a j 
■1 

TT :| 
22 1 



differently" . "I believe the customers preferred the method which RCA 
had been following and which RCA elected to continue." (Tr. 11206-09.) 

NCR went some distance in the direction of unbundling. 
On October 1, 1969, it stated its belief "that each user of its 
computer systems must be provided with a certain essential amount 
of software, systems support, and educational services if he is to 
successfully install the system and begin to benefit from his 
investment. NCR believes that this basic package of supporting 
services must be the responsibility of the equipment manufacturer." 
(DX 346, p. 1, emphasis in original.)* NCR recognized that there 



* NCR expressed the view in its 1969 report to stockholders that, 

"The deluge of new concepts and new equipment which has 
flooded the information processing industry in recent years 
points up dramatically the need for ever-greater customer support. 



•" j "Indeed, the growth of the industry will continue to depend 

\ in large measure upon its ability, through supporting services, 

2- I to adapt these new concepts and equipment to the requirements 

.1 of different organizations. Thus, during the 1970s increasing 

25 I funds and effort will be devoted toward broadening the spectrum 

i of customer assistance." (DX 367, pp. 19-20.) 



-469- 



J 



J 1 



would be considerable variance in the level of support required by 
different customers and stated that "it will continue to be NCR's 
policy to provide, as part of the basic hardware price, that amount 
^ | of software and support which will realistically insure that a 
prudent user will be able to install and successfully utilize his 
NCR computer system". An allowance, based on the size of the 
system amounting to "approximately 30 man-days of support for each 
$1,000 of monthly rental" was to be provided with support above 
that level billed separately. The same principle was to apply to 
educational support and software "including both applied programs 
and computer languages". (Hangen, Tr. 10721-24; DX 346.) 



2 | On January 1, 1970, however, NCR announced a change in 



3 

c 



its policy stating: 

"After further evaluation, it has been decided not to 
price all basic and applied software and not to establish an 
allowance against which such chargeable software would be 
applied. The NCR software pricing plan will be to continue to 
establish pricing for software products on a selective basis, 
considering the value to the customer, uniqueness, and other 
factors." (DX 386, p. 2, emphasis in original.) 



•t 
■I 



si 



rr 1 



i 
>2 1 



There was much less disagreement in 1969 and 1970 on the 
question of whether or not operating systems or systems control 
i programming should be unbundled.* IBM did not unbundle such pro- 
gramming, stating: "System control programming is an essential part 



" 1 * By the early 1970s only CDC had unbundled its operating system. 

1 (Norris, Tr. 5647; Goetz, Tr. 17530.) 
1*. \ 



-470- 



L 
Z 
3 

4* 

a 
T 
3 ; 

i 
i 

ia i 

t 
I 

U ! 
12 

T 



of a data processing system. It is fundamental to the operation and 

maintenance of the system and will be made available as part of the 

system," (PX 3351, p. 4; see also PX 2454, p. 1; PX 3352, p. 5.) 

The fact that operating systems were essential was widely 
v 
recognized, Enfield testified that he did not "see how" a supplier of 

computer systems could market its equipment without making available 

some form of an operating system, at least following the introduction 

of System/360, by either producing the operating system itself or 

arranging for it from some external source. (Tr. 20740-41, 21074.) 

Welke testified to the same effect, (Tr. 19223.) Dr. Perils of Yale 

testified that 



LZ 



"operating systems are . . . indigenous to all major computers 
at the present time. They manage the computer resources and 
they really could be part of the hardware except that their 
functions are not well enough understood at the present time 
ti • to make it economically feasible to put it into hardware". 
(Tr. 1344.) 



IS 



According to Perlis, operating systems are "crucial to the 
successful operation of almost every computer around today". 
u ■ (Tr. 1348.) As a result, operating systems are "typically" designed 



to ; 

•* it for a particular fit with a particular computer, because "[t]hey 

i9 | depend very strongly on the particular resources and the way those 

; i 

& j resources are organized in a particular machine, and they do this 

:l 
~ ; so that they can achieve the most efficient operation possible . . . 

22 :j to take maximum advantage of the idiosyncrasies of the hardware". 



** j (Perlis, Tr. 1986.) Thus, Withington testified that hardware and 

2- I software "are now necessarily designed as one, designed to execute 

25 j f rom the same architecture". (Tr. 55919-20; see also DX 491, p. 5.) 

1 

I -471- 



3 



Hence, separate pricing of operating systems would require "arbitrary 
allocations".. (Withington, Tr. 56798.) Indeed, Withington wrote in 
June 1969, shortly before IBM's June 23 unbundling announcement, that 

■ systems software was "essential to the operation of modern computers 
and is designed contemporaneously with the machines. It is not 

;j possible to separate its development costs from those of the computers 

themselves, nor is it possible for the machine to operate without some 

version of the operating system" . He concluded that this was a 

"complex area" and that "basic skeletons of the operating systems" 

were, at that time, "likely to be provided free with every machine 

. . . because there is no rational way to separate them". (PX 4834, 

p. 11.) 

Ray Macdonald of Burroughs testified: 

"[W]e had extensive discussions on systems software, and I 
believe that our conclusion after some experimentation, and 
quite a bit of back and forth debate, was that the systems 
software that I have described is in fact an inseparable 
part of the system for the average user. 



m ' "Now there may be the very unusual user which represents 

r -j an extremely small portion of the total market who may . . . 

have the sophistication to consider a different mix or differ- 

5 . ent system software for his own purposes, but first of all, I 

j think this is a very, very small portion of the total market, 

* : j and certainly not suitable for the vast majority of the market." 

; (Tr. 6977-78.) 
3 



i 

T ;l 



Thus, as the 1960s ended IBM had embarked on a course of 



4 separately pricing certain of its software and services . As we shall 
| see, IBM continued on this course during the 1970s, separately pricing 
| increasing amounts of its software and services in response to rapidly 

4 i 

! changing market requirements and technological advance. 
! -472- 



■l _. . - , , „ „ i 



14- ; 



. - it 



1/ • 



(DX 10; Eckert, Tr. 1014-17), Univac had manufactured and was still 
marketing several incompatible product lines (represented in 1964 by 
the 490, the Univac III and the 1107) , each requiring different soft- 
j ware. Moreover, Univac had failed to provide successors to its 
obsolete products. (See PX 4829, p. 20; DX 8, pp. 1-2; DX 10; DX 14, 



' In 1964, "after it had become apparent to the rest of the 



1S:| 

..p. 1.) 
19 i 

za :| 

1 industry that magnetic disks were superior" , Sperry was still marketing 

.j its FASTRAND drum instead of quickly proceeding with disk development, 
22 'I 



■I * R. E. McDonald was President of the Univac Division from 1966 
24,; to 1971. (McDonald, Tr. 2769, 2776-78.) 

ys i ** J. P. Eckert was a Vice President of the Univac Division and 
'technical advisor to the President of Sperry Rand at the time of his 
:! testimony, having held that position from "about 1960". (Eckert, 
; {Tr. 710.) 

'! 

i| -473- 



40. Sperry Rand/Univac. Although Sperry Rand's Univac Divi- j 
sion entered the second half of the 1960s lagging substantially behind 
the industry leaders in the areas of product compatibility and storage 
technology, it was able, by the end of the decade, to reestablish 
itself as a major' force in EDP, logging substantial gains in revenues, 
organization and technology. 

a. Univac* s Problems in 1964. Univac, in 1964, was in a 
state of some disarray * It was in the midst of a succession of 
presidents (Eckert, Tr. 1008-13r McDonald, Tr. 3785-88)* and was "still 
suffering" from the "great drawback" of its "inability to assemble 
a smoothly working, reasonably permanent management team". (PX 4829, 
p. 20.) Additionally, despite the suggestion of Dr. Eckert** 



i 

5 

3 

7\ 
3 

2 

a 
z 

.6 
J 
,S 

19 

za 



a delay which had a substantial adverse effect on the marketing of its 
computer systems. (Withington, Tr. 56455, 56485-87.) Consequently, 
Univac was compelled to purchase disks from other suppliers " for a 
while".* (Withington-, Tr. 56243-44.) 

Univac'.s financial results during the first half of the 
decade were not particularly encouraging. In 1962, the corporation 
had found that "the rate of technological obsolescence" required it 
. to write down the value of its older EDP equipment by more than $50 
million, and to accelerate the depreciation of its newer models. 
(DX 69, p. 3.) In 1964, Univac was "losing money" (McDonald, Tr. 3813) 
and experiencing a relatively slow rate of revenue growth. (14.8% 
compound growth rate from 1960-64 compared to 27.4% from 1956 to 
1960.) (See DX 8224, p. 624.) 

Notwithstanding its limited success , Sperry reported to its 

shareholders in its 1965 Annual Report: 

"Data processing is a dynamic industry, having great growth 
potential. It has established a place in the world's economy 

j that is essential and will continue to grow. Such dependence 

! upon any industry in the past has not only led to growth but 

also profitability. Therefore, we have determined that we will 

! remain in and grow with the data processing business." (DX 13983, 

I P. 6.) 



i * IBM employees reported, in 1968, that suppliers of disk drives 

21 i to Univac included Vermont Research, Bryant, Data Disc and Memorex. 
: i (PX 2267B, p» 27.) Univac continued to purchase disk drives from 

22 :j Memorex through 1970 (Guzy, Tr. 33170-71), and from CalComp through I 
| 1973. (PX 5584, p. 16.) It also purchased disk drives from Peripheral ! 

22 | Systems Corp. in 1969. (DX 1302, p. 1.) ! 



i 

1 -474- 



L i 



2 i Univac concentrated its marketing efforts on the Federal government and 



5 ! 

I 
5 i 

7; 

i 

3 j 

i 

i 

IQ ! 

! 

i 

II : 

i 

IZj 
13 ; 



According to McDonald, during the period from 196 3 to 1971 



airline reservations users. (Tr. 2890-91.) The Federal government was 

a very important customer for Univac in. the 1960s, as it was for most 

of the industry; Sperry reported that in the fiscal year which ended 

March 31, 1964, for example, the Navy had ordered four 490 systems for 

world-wide inventory control, the Marine Corps had ordered three Univac 

III systems far similar applications and the Air Force had "ordered more 
than 150 UNIVAC 1050-11 systems, as well as three UNIVAC 1107' s for 

logistic control purposes". (DX 13913, p. 16.) The Air Force order 
alone was, as Withington noted, "large enough to cause a bulge in ship- 
ment statistics". (PX 4830, p. 22.) During the 1960s Univac claimed 
"a complete array of computers" for the military. In its 1964 Annual 
Report, Sperry Rand contended that "no other company in the industry 
[could] match this range". (DX 13913, p. 12.) Several computer systems 



were offered by Sperry Rand to satisfy shipboard, airborne, van-mounted 
'land aerospace military and space requirements.* Univac was the prime 
'{supplier of the militarized AN/UYK-5 & 7** which were the standard 
milspec computers for the U. S. Navy. Univac also had a broad range of 



17 
13 

15 ; 
•i 

20 ] 

'! * Sperry Rand also had large commitments to the space program. 

Zl iFor example, eleven Univac 1218 systems were to be delivered in 1964 to 
|NASA tracking sites to assist in tracking the Gemini Capsule. A Univac 

22 .1218 was selected for the "mobile-wing, limited-warfare intelligence 
[complex". (DX 13913, p. 12.) 

. j ** The AN/UYK 7, a chird-generation computer, used a general 
2± '.purpose software package called Gipsy, developed by the Naval Electro- 
nics Laboratory in San Diego. Gipsy provided the capabilities of a 
?= master control and data base handling program with a maximum degree 
'of hardware independence. (DX 5117, p. 1.) 

.1 

: i -475- 



computers oriented to Navy milspec requirements which were in popular 
use aboard. Navy vessels, performing a wide variety of applications. 
For example, the Naval Tactical Data System (NTDS) application aboard 

• the U. S. S. Enterprise had a Univac USQ20 (1206) as their central 
computers. (DX 69, pp. 12-14.) In addition, the AN/UYK-5 computers 
were used on Navy vessels to process maintenance records, supply and 
accounting applications . (DX 5123, p. 3.) The Marine Corps used the 
Sperry Rand 1005 systems (AN/UYK-5 [V] ) for their field van-mounted 
applications.. Van-mounted 1005s were also used by the Army in their 
PERMACAPS and DLOGS systems in Germany, Vietnam, Korea and around the 
United States. (DX 5410, Fullerton, pp. 36-37.) At the White Sands 
Missile Range (WSMR) in New Mexico the Sperry Rand 1218* was used for 
a variety of applications including: missile guidance and tracking, 
data reduction and analysis, simulation, communications, logistics 
management, and satellite tracking. (Plaintiff's Admissions, Set II, 
H 768. 0-. 4.) The Univac 1108 at WSMR was also used for missile guidance 

; j and tracking. ( Id. , 1f 748.2-. 4. ) 

7 ii 

j The Naval Electronics Laboratory in San Diego acquired: 

3 1 

:| 1 IBM 360/65; 1 CDC 8090; 1 Sperry Rand 1230; 5 Sperry Rand USQ20s 
■9 j 

I (CP 642A/B) ; and 1 Sperry Rand AN/UYK-7 (CP 890.) (Id., <i 702.0.) As 

» ] 

i of 1974, applications of a general data processing nature previously 

- i • 

\ processed on the Univac CP-667, USQ-17 and USQ-20 computers were to 

be transferred to the IBM Svstem/360 Model 65 along with those that 
23 ! 

! had been run on the two CDC 1604s. ( Id. , If 702.15.) 

14. i 



f 

L 
t 
x 

r 
i 
i 

j 

l ; 
Z ! 
3 
4- 



I * The militarized 1218 and the commercial UNIVAC 418 are identical 
■! in design and the mainframes do not vary at all. (DX 90 8 8.) 



-476- 



Li 
Z 

z 

3 

a 

7 

3 

*! 
10 
Il f 
12 
13 
L*. 
15 

Iff 

17 
IS 
19 

'I 
li 

22:1 



The Department of Navy reported the use of Sperry Rand 
AN/USQ-20s, AN/UYK-5s, Sperry Rand 1219s, and a Sperry Rand AN/USQ-17 
for the Navy Management Information System for Education and Training. 
(DX 2992, pp. 592-93, 1123.) 

In the area of" airline reservations, British European 
Airways ordered a 490 in 1964 (DX 13913, p. 13); two years later, in 
fiscal 1966, Univac reported that it had been awarded the "biggest com- 
mercial computer contract ever awarded", a $39 million contract from 
United Airlines, "to design and build a computerized information system 
that [would] handle United 's needs through 1975". (DX 61, p. 9.) As 
it turned out, Univac was unsuccessful in its bid to meet United' s 
requirements, and the effort was "aborted" in 1970, with United 
Airlines moving to an IBM system.* (O'Neill, Tr. 76015-17, 76231-32.) 

b. The 1108. The United Airlines system was to have been 
"based on Univac 's 1108 f s". (O'Neill, Tr. 76231.) This computer, 
introduced in 1964 (DX 13983, p. 14), was compatible with the thin- 
film 1107, and was intended for Univac' s "large-scale users". (DX 
14, p. 1.) Withington viewed the 1108 as "technically impressive", 
claiming that its "very fast control memory" marked "the first signi- 
ficant appearance of integrated circuits in commercial computers . . 
. ." (PX 4829, p. 20.) The 1108-11, a "time-shared version" of the 



i * "The reason that United decided to terminate that activity was 
23 that they concluded that the system being developed [for] United at 

{ Univac would not accommodate their projected volume. They subsequently 
24. : { installed IBM 360/65S, and later installed IBM 195s for their 

.{ passenger service system." (O'Neill, Tr. 76016-17.) 



re 



-477- 



u 

f 

T 
I 
* 

2 

I 

Z\ 

3 

£; 

S 

5 

7 

,3 

3 



1108 was introduced in 1965. (PX 4830, p. 22.) The 1108 was not 
delivered in volume until late 1966. (PX 4832, p. 18.) By 1967, the 
1108 "accounted, for about half the value of Sperry Rand's shipments". 
(PX 4833, p. 17.) 

Univac- continued to develop and extend its 1108 system 
and related machines through the late 1960s. Univac, in fiscal 1969, 
announced the 1106; "a smaller, compatible version of the 1108 system". 
(DX 3271, p. 5.) In 1967, Univac entered "the data services field with 
a service bureau network Of 1108 's directly connected to small computers 
on users 1 premises". (PX 4833, p. 17.) 

Univac 1108s were employed in a wide variety of commercial 
contexts. The 1968 Sperry Annual Report showed a picture of the 
Univac 1108 scheduling trains for the French National Railway. 
|| (DX 13914, p. 5.) In addition to United' s reservation system, Fuji 
Bank Ltd. , Tokyo, in 1969 inaugurated a nationwide on-line banking 
system using an 1108, according to the Sperry Rand 1969 Annual 
Report (DX 3271, p. 7) ; and the Sun Oil Company ordered an 1108 
system in 1968 for use in processing business and scientific problems 
! (DX 13914, p. 16), to name but a few examples. As Sperry management 
noted in its 1970 report: 



» ! 



" . . . The Univac large-scale computer systems — 



£ j especially the 1100 series— are acknowledged to be the most 

"" • versatile processors available. The UNIVAC 1108 and 1106 

?2 "i systems, in addition to having unparalleled capability for 

j scientific and engineering applications , have gained wide 

?3 ; acceptance among commercial/industrial users for business 



i 

i 

2* I 



;; -478- 



I 

2 

a 

4- 

a 

m 

7 1 
3 

IQ 

II 

12 
12 
14 
15 
15 
17 
IS 



data processing and communications tasks." (DX 13915, p. 5.)* 
In March 1969, Sperry Rand management reported that "the 
backlog for Univac 1108 computer systems continues at a high 
level. ... It provides the Company with an entree into the 
market for management information systems because of the computer's 
communications and multiprocessing capabilities in both business 
and scientific applications". (DX 3271, p. 5.) 

Development of the 1108 was not without its problems, 
however,- the 1108 operating system, EXEC- VI II, had "major problems 
in its initial stages". (J. Jones, Tr. 79631; PX 4834, p. 25.) These 
problems, similar to those encountered by other manufacturers with 
complex operating systems during the 1960s (see Perlis , Tr. 2002-03; 
Weil, Tr. 7217-19; McCollister, Tr. 9694-97; Rooney, Tr. 12132-36, 
; 12349-50; Conrad, Tr. 14088-89; Withington, Tr. 56727-31), came 
relatively later for Sperry Rand "because it was not attempting to 
offer systems programs as complex and advanced as the other competitors 
were". (Withington, Tr. 56736.) During the late 1960s, Univac failed - 
to deliver operating systems which completely met their advertised 



» :| i 

;( * NASA's Marshall Space Flight Center (MSFC) Computation Laboratory ; 

20 | utilized 1108s to perform both "scientific data processing**' and j 
i "administrative data processing". (Plaintiff's Admissions, Set IV, \ 

21 -j if- 386.0.) The Slidell Computer Complex at Marshall also used two ] 
J1108 systems for rocket stage design work, scientific applications j 

22 land "some administrative data processing".. (Plaintiff's Admissions, j 

Set IV, flK 390.0, 392.0, 394.1, 401.0.) Five Univac 1108s were j 

23 (installed at the White Sands Missile Range (WSMR) , utilized by the I 
[Army, Navy, and Air Force: two of the 1108s are employed in real- 

24 Itime missile performance computations; two others provide back-up, 
j batch processing of test data and remote time-sharing ability; and 

25 jthe fifth is used for batch processing of classified' data. (Plain- \ 
i tiff's Admissions, Set II, "[fl 746.4, 748. 0-. 7.) 

1 I 

'! -479- 



capabilities and, indeed, EXEC- VI I I was delayed at least two or three ' 
years , not meeting its. advertised capabilities until sometime in the 
early 1970s.*" (Perils-, Tr. 2003; Wlthington, Tr. 56737.) 

c. The Product Line Task Force. The 1108, though successful, 
was not an answer to Univae ' s need for a compatible product line. As 
we have seen, it was announced at approximately the same time that 
Eckert, in his capacity as head of the Gemini Committee, was calling 
for unification of Univae' s dissimilar product lines. In 1965, in 
the wake of IBM's System/360 announcement, Frank Forster, Univae' s 
President from July 1964 to early 1966 (DX 13983, p. 6r DX 61, 
pp. 2-3) , set up a Product Line Task Force to review Univae products 
and to help him make decisions about their future. (McDonald, Tr. 
3804-05; see also DX 13.) 

The task force, in February 1965, reported that it believed 
that Univae 's manufacturing costs were higher than those of IBM, and 
that : 



i 

r\ 
i 

? 

3 ; 

i 

z 

A 

.4- 

6 r 

• "IBM's heavy investment in product research is beginning 

7 f to bear fruit. Its developments in circuits, microprogramming 

| techniques, memories, and mass storage suggest that for the 

3 :| first time in the short history of the industry, IBM has 

•| acquired a definite technological leadership; this, together 

,3 ;; with our cost situation, may leave us little to sell. 

; (DX 15, p. 2.) 

:g: i 

i In its next report, issued in March 1965, the task force observed that 

SI ! • 

■] both Honeywell and RCA had committed themselves to the production of 

;j integrated computer families (the Honeywell 200 series and RCA's 
5 i 



Z4 i 

;| * As a result of the delay in developing EXEC-VIII, NASA, for 

;e j example, was able to renegotiate its contract with Univae to 

J include the grant of free computer time as a "slippage" penalty. 

;j (DX 5654, pp. 114-15,-231-32.) 

;! 

I -480- 



Li 

Z 

3 

4- 

7 

a 

3- 

ia 

u 

12 

n 

14- 

15 i 



Spectra 70 line) in the "tailwind created by . . . IBM". (DX 16, 

p. 2.) The report quoted the editor of Datamation ; 

"U&TW&C is the big question mark * . • every month until a 
new line is announced weakens their chances of success . . . 
and it's not clear they'll offer a complete line at all. 
Anything less- could, relegate them to the second division." 
(Feb.. r 1*65, p. 8$.) ( Id. ) 

Nonetheless, the task force was unsure whether Univac should try 

to match IBM's 360 or take some other action. Specifically, it 

expressed the concern 

"that the RCA and Honeywell moves, although based on clever 
sales strategies, may not make such good sense financially. 
Both are based on the assumption that now that IBM has made 
its move, the pace of obsolescence will slow down, and longer 
writeoffs will be possible than in the past. It is our 
opinion that in about five years this assumption will prove to 
be catastrophic to anyone who bases his product line on it." 
(Id.) 



Ultimately, Univac decided not to introduce a full 
spectrum product line but to introduce only three machines , called 
models A, B and C. In consonance with its "concern" about future 
^° I technological developments rendering obsolete an entire product 
if i line, the task force called for accelerating development of the 
-S ; t model at the low end of the line, the model A, which was to be a 
15 |j 360-compatible processor targeted between the 360/20 and 360/30, to 

20 ;> take advantage of the "large and barely exploited market for a low- 
i) 

21 :j priced scientific computer". ( Id. , pp. 2-3.) The task force 
] 

22 ;} observed, however, that: 

22 •} "The announcement of Model A will have an effect on 

I the whole product line, all the way up to the 110 8A. Regard- 
2^ j less of what is claimed, the fact that model A contains the 
360 repertoire will tell the world that our other products 
may be dead ends". (Id. , p. 5.) 



-481- 



i 

*•! 
i 

3 

7 
3 

Q 

* 

2 
3 

.4 
5 
,3 
.7 



d» The £000. Series . The task force had been convened 

to consider Univac' s product strategy nearly three years after 

IBM's SPREAD Committee report,- its reports appeared nearly a year 

after the announcement of System/ 360. (DX 16, p. 1.) Univac 

finally announced its third generation compatible computer family, 

the 9000 series (corresponding to the previously mentioned models 

A, B and C) in the spring of 1966. Called a "line of small and 

medium-sized computer systems", Univac' s initial offering included 

"the 9200, a low-cost, internally programmed punch-card system, and 

the 9300, a high-performance card and tape system". (DX 70, p. 9.) 

While the 9000 "aimed at compatibility" with IBM's 360 (Eckert, Tr. 

908) , it was not truly compatible: 

"[A] new line, compatible with IBM 360 coding . . . would have 
probably solved the problem. While the 9200, 9300 and 9400 
are IBM like in, their order code, they are not enough alike 
to do us any real good. We have had loads of people prove 
to us why we can't be IBM compatible and very little real 
effort to be IBM compatible, either in our software or our 
hardware efforts." (DX 10, p. 1; see also McDonald, Tr. 
3803-04.) 

The 9000 series was upward but not downward compatible 



•* ,j among the three models. Thus, "if a person had programmed something 

— ;j for some of these smaller machines he could use it in one of the 
t 

"6 ;j larger machines but not the other way around." (Eckert, Tr. 906- 

i 

" ] 07.) It also was not compatible with the 1100 series. (Eckert, Tr. 

4 
22 j 908.) 

- I 

'-* j The third machine of the line, the Model 9400, was first 

i- ! announced in January 1968 (DX 13914, p. 6) , and delivered in 1970 
Zz j "from factories in the United States, West Germany and Japan". (DX 

! 

■! -482- 



i 

I; 

i 

f I 
i 

! 



3271, p. 7.) The 9000 series was intended to "enable smaller 
companies to benefit from the advantages of computer power. . . . 
Typical customers [were] a savings and loan association in Kansas 
City, an aviation company in California and a wholesale grocer near 
Philadelphia. " (DX 13914, p. 16.) 

Univac both manufactured its own peripherals and purchased 



7 ij peripherals from others, remarketing them as part of its computer 



S 

9 

10 

11 
12 
13 
14- 



systems.* For a short period, it marketed to other manufacturers 
its peripheral devices which in turn were remarketed as part of 
other systems.. (McDonald, Tr. 4053-55.) Further, its own products 
were used as part of systems in another way. The 9000 series, for 
example, were sometimes used as terminals to other manufacturers' 
systems. (McDonald, Tr. 3969? Withington, Tr. 56981.) As a 1970 
jl Univac advertisement said: 



q. -J "They are widely used as either central site systems or 

terminal systems. As terminals they may be upgraded, without 
+ s >.i reprogramming , in low-cost steps to grow with your processing 
lS 'l needs." (DX 13939 , p. 176.) 

U jl In addition to acquiring peripherals from other manu- 

-S ; f acturers , Univac contracted with software houses to have work done 

*s : j when it did not have sufficient in-house capability to meet its 

:| 
20 j requirements and did not wish to expand internally to meet a peak 

t 

T5 H 



* For example, Univac purchased tape drives from Ampex (Ashbridge, 
23 | Tr. 34851) and disk drives from Memorex, Calcomp and Peripheral Systems 

j (Guzy, Tr. 33170-71; PX 5584, p. 16; DX 1302, p. 1.) In 1968, IBM 
24- i employees reported that Univac also purchased disk drives from Vermont 

J Research, Bryant and Data Disc and tape drives from Potter Instruments 
25 | and OKI. (PX 2267B, p. 27.) 

i 

I -483- 



load. (Eckert, Tr. 915-16.) McDonald testified that Univac purchased 
"software assistance from the Computer Sciences Corporation and also 
from University Computer Company". (Tr. 4024.) 

Univac- both leased and sold its EDP equipment. McDonald 
rewrote in 1967 thatt 



I : 

jj 
Ti 

T. 
Z 

m 

4, 
5 
6 

7 i: 
s 



M [approximately 50 per cent of the Division's products are 
sold outright with the remainder leased by customers on a one- 
year to five-year basis." (PX 1, p. 3.) 

Univac provided support services to its customers as well. (McDonald, 

Tr. 2893-96.) McDonald testified that Univac had to provide these 

services if it "were to compete successfully", since IBM did so. 

(Tr. 2895-96.) However, Univac did not unbundle when IBM did in 1969, 

because: 

"[W]e felt that there would be considerable anxiety in the 
marketplace . . . and we felt that it would be to our 
competitive advantage to maintain our previous pricing 
policy . . . and I think this was effective, at least for 
a period of time." (McDonald, Tr. 2896.) 

McDonald testified that Univac' s pricing policy between 

1963 and 1971 was "to provide the potential customer with a system 

that would perform his requirements at a price that would generally 

be 10 percent, as a rule of thumb, below the price offered by IBM", 



:{ not taking into account the performance of associated peripheral 
.1 devices. (Tr. 2883-84; 4190-91.) Univac attempted to set its 

~ :! • 

\ products' price/performance between IBM's products, much as RCA had 
'? i 

^ : |done with its Spectra series. (McDonald, Tr. 4132-83.) Consider- 

ing that Univac 's 9000 series was announced two years after Sys- 
2- i 

jtem/360, Univac' s pricing approach was perfectly understandable. 

IBM was not the only competitor about which Univac was 



25 



i -484- 

i 



1 



3 _;■ industry" [IBM, Univac, CDC, RCA, G£, Honeywell, Burroughs, and NCR] 



concerned, however* While McDonald, in 1967, identified "eight 
major hardware manufacturers" who were "[a]t the hard core of the 



(PX 1,'pp*. 6, 12; see also McDonald, Tr. 28:04-06), he recognized that: 

"[b]y the 19'60's, there were up to 50 major suppliers of 
automatic computing digital and analog computers and data 
processors. Over 700 organizations with some 30,000 
persons were engaged in one part or another of the computer 
field." (PX 1, p.. 1.) 

These included peripheral manufacturers, software suppliers, service 

centers, and leasing companies. ( Id. , p. 12.) 

In the middle 1960s Univac management became "concerned" 

about leasing companies. Forster wrote to McDonald in 1966, stating 

that he had: 

"some apprehension and also some prejudice in that I 
consider them to be parasitic. ... If computers do 
not stay on rental, since they have no loyalty to any 
particular equipment their manner of disposal could be 
damaging." (DX 78.) 

Univac 1 s concern about the "manner of disposal" of leasing company 

■ equipment was that the leasing company would at some later time 

market it at very low prices, in effect "dumping it" on the market, 

knocking Univac' s own equipment out of customer installations. 



4-1 

i 

» 
i 

t\ 

8 [ 

5 i 

10 ; 
III 
121 
13 

14- 1 

15 

IS 

1? ; 

I8 j; 

id 

j (McDonald, Tr. 4017; DX 76.) 

20! 

' Univac responded to this concern. In January 1969, 

21 jj 

] management approved revisions in Univac f s long-term lease plan which 

:[ were designed to "decrease future vulnerability" to third-party 

I leasing companies and which included the adoption of step-down 

24 ! 

. payment plans for long-term leases and price-cutting of five-year 

I lease rates for Univac ' s "most profitable systems". (McDonald, 
•\ 

II -485- 



i il 
w 

3 
T 
3 

3 

a 

Z 
A 

£ ! 

LS 



Tr. 3988-; DX 76, p. 5.) 

e. Univac 's Success in the Late 1960s. Despite the fact 
that Univac did not offer a single compatible family with the 
breadth and compatibility of , the IBM 360, it experienced substan- 
: tial growth in its EDP business during that time. At the end of 
1965, prior to volume shipments of the 1108 or the announcement of 
the 9000 Series, Univac 's U.S. EDP revenues were $203 million; as 
of year end 1970, its U.S. EDP revenues were $478 million. (DX 
8224, p. 624.) McDonald estimated in 1974 that "revenue growth 
since fiscal 1965 [had] been 284%, or a 16% compound annual growth 
rate — in excess of that of the computer industry as a whole". 
(McDonald, Tr. 3867-68; DX 71, p. 7.) By fiscal 1969, the Univac 
division had become "the largest contributor to [Sperry's] revenues 
and earnings".* (DX 3271, p. 2.) 

Univac' s growth was not limited to the United States. 
From at least the 1960s onward Univac offered a single worldwide 
product line. (Withington, Tr. 57602-03.) Thus, in 1967 Univac' s 
International Division conducted operations through 32 subsidiaries 
•J and distributors in Canada, Central and South America, Europe and 
'} the Far East. Sales and service offices were situated in Belgium, 

(England, France, Germany, Italy, the Netherlands, Norway, Sweden, 

21 ] ■ 

j Switzerland, Australia, Brazil, Canada, Mexico, Argentina, Colombia 

22 1 



* Withington echoed the turn-around: 



24 



"The Univac Division became the largest single contributor to 
?e 'I the profits of the corporation (it seems only a short time 
~ I ago that Univac was castigated as the largest single drain 
J on them!)." (PX 4834, p. 24.) 



-486- 



Li 
Z 

3 

5 
7 
3 

9 

10 

II 



and Venezuela. (PX 1, p. 4; see also McDonald, Tr. 3839-42.) 

McDonald predicted in 1967 that "overseas markets will grow at a 

more rapid rate than that of domestic markets .... DNIVAC ser* 

; iously intends to participate in the rapidly developing European 

^ market'* and **wili routinely work across many international boundaries". 

(PX 1,- pp. 5-6.) For 1970 Univac reported that its "international 

t business [was] growing at an even higher rate than the domestic 

operations". (DX 13915 r p* 7.) 

Univac made great strides in the last half of the 1960s 

despite its slow start in undertaking a compatible family of products 

and its reluctance to accept disk technology. McDonald recognized 

.-jj what the problem had been and what would be required to solve it: 

£2 ,1 "Planning will be a requisite to survival on the basis 

upon which profitable business development can be struc- 
T ^ ; . tured. . . . The combined magnitude of both opportunity 
i and risk superimposed upon the rapidly changing pace of 

the industry will rule out success based upon 'seat of the 
pants' decision-making. The old technique of fumble and 
correct errors is out. There will not be time in the 
future to recover from serious mistakes without suffering 
;J severe penalties. We, therefore, must measure daily 
17 ;| events against a flexible, preconceived plan of action 

j in order to react in a timely fashion, competitively. 
-S :[ Hard planning will be a part of daily activity. It will 
not be a luxury in the future. 



io 



IS i| 



:| "This is the precise area of one of UNIVAC s greatest 

20 :j past weaknesses. It is an area which has received con- 
centrated attention since 1964 and will continue to receive 
emphasis in the future." (PX 1, p. 7.) 






22 .1 Finally, Univac was back on its way to becoming a successful 

22 j computer company . 

2*1 



! -487- 



41. General Electric. At the time of IBM's announcement 
of System/360/ General Electric was (as it still is) a large corpora- 
tion. From, that time to the end of the 1960s, it was always in the 
top six of the Fortune 500. ,(R. Jones, Tr. 8754.) Its corporate -wide 

ii revenue grew from $5.1 billion in 1964 to $8.4 billion in 1969. 

i{ 
t,| (DX 13667, p.. 1; DX 556, n. 2.) IBM's corporate revenue in 1964 was 

r!| $3.2 billion (PX 5771, p. 3) and $7.2 billion in 1969. (DX 3364, 
j i p. 5.) GE was larger than IBM throughout the entire period. (DX 556, 
p. 28? DX 3364, pp. 59-60.) However, whereas most of IBM's domestic 
revenue during the period 1964-1969 came from its EDP business (see 
DX 3811; DX 3364, pp. 47, 48, 53, 54; PX 5771, pp. 32, 36; DX 13677, 
pp. 33, 37; DX 13678, pp. 33, 37; DX 13679, pp. 33, 37; DX 13680, 
pp. 45, 46, 53, 54), virtually none of GE ' s did. As the chart below 
Ishows, at no time during the period 1964-1969 was GE ' s U. S. EDP revenue 
9 \ more than 3-1/2% of its total U. S. revenue. (See Weil, Tr. 7260.)* 

■0 

! 

si 

7 
S 



* 

1 

L ; 
1 
I 



5 ;j * Plaintiff called four witnesses who testified about the GE 

] computer business. They were John W. Weil, who was in GE's computer 

6 ;| business from 1963 through 1970, as Manager of Engineering 
:i from 1964 through 1966 and Manager of Advanced Systems and 

LI .!• Technology Operation thereafter (Weil, Tr. 7003/7007-08, 7072); 

i John L. Ingersoll, who was involved with GE's computer business from 
'2. :l 1967 to 1970 as a financial manager and a staff member of the Ventures 

| Task Force (Ingersoll, Tr. 8042-43, 8097) ; Richard M. Bloch, who was 
2 | Manager of the Advanced Systems Division of GE from November 196 8 to 

; mid-1971 (R. Bloch, Tr. 7615-16, 7755, 7777); and Reginald H. Jones, 
£4 j who held top management positions at GE beginning in 1961 and became 

! Chairman of the Board in December 1972. (R. Jones, Tr. 8752-53.) 



-488- 



Li 
z\ 
z \ 

1 

4-i 
5 ! 
6 
7 
3 
9 
10 

ii i 

12 
13 
14 
15 
16 
17 
IS 
19 
20 



GE Total GE 

U.S. U.S. EDP 

Year Revenue Revenue 

(in millions) 

1964 $4011.5 $ 53.4 

1965 4952.6 66.5 

1966 5698.3 99.0 

1967 6129.2 143.1 

1968 6664.6 180.0 

1969 6638.0 219.6 

(PX 326 (DX 13668, pp. B, 3); PX 327, pp. 2, 36; PX 328, pp. 2, 23; 
DX 556, pp. 2, 18; DX 8224, o~ 6.; DX 8631, pp. 31, 37; DX 13667, pp.. 
B, 14; DX 13669, pp. 3, 4; DX 14484, p. Rl; DX 7320.) 



% GE 


U. 


S. 


EDP 


to 


GE 


Total U.S. 




1. 


3 




1. 


4 




1. 


.7 




2. 


.3 




2. 


.7. 




3 


.3 



In 1963, computers were a part of the "industrial compo- 
nents and materials area" at GE which accounted for 28% of GE's 
revenues in 196 3. That area also included advanced controls for 
machine tools, Lexan plastics, silicone chemicals, component motors, 
appliance controls and lamp ballasts. The remainder of GE's business 
j was derived from consumer goods (26% of revenue) , including appliances 
! television, and lamps, among others; heavy capital goods (24%) , 



i including diesel electric locomotives and power generating and 
21 il 



22 

23 
24 
25 



transmitting equipment; and defense sales (22%) , including jet 
engines and missile guidance systems. (PX 325, p. 10.) 

Notwithstanding the small part played by computers in the 
GE hierarchy, GE had to be considered one of the most significant of 
I3M's competitors in the computer industry in the 1960s because, as 

-489- 



L ; Richard M. Bloch, who joined GE in 1968 as the Manager of the 

;f 
I ''■ Advanced Systems Division, testified, GE "was probably the greatest 

l\ electrical and electronic technical organization, technically 

M" oriented organization in the world, and with very strong financial 

f | ■ resources " . (R. Bloch, Tr. 7615-17.) Similarly, in 1964 Withington 

i : 
5- ; wrote:. "GE's long— term potential must be considered greater than 

j\ that of any IBM competitor". (PX 4829, p. 19.) John W. Weil, who 

j i) was the manager of engineering for GE's Computer Department from 

196 4 to 1966 and thereafter the Manager of GE's Advanced Systems 

and Technology Operation until 1970 , testified that he believed 

that "GE had the resources and technological capability to become a 

major force in the computer industry". (Weil, Tr. 7007-08, 7072, 7173.) 

With all that technological potential and financial power, GE was 

called the "sleeping giant". (R. Bloch, Tr. 7788-89; PX 353, p. 43.) 

But, in the computer field at least, the "sleeping giant" 

never woke up. Its efforts in computers in the 1960s ended with the 

sale of most of its computer business to Honeywell in the merger 

that created Honeywell Information Systems. The story of how GE 

failed to capitalize on its advantages and succeed is the story of 

lack of corporate commitment, inadequate management and a failure to 

keep up with the demands of the, market as technology and competition 

advanced . 

a. The GE 400 Series. During the year 196 3, GE was 



l\ from NCR) and the 225. (Weil, Tr. 7005-06; see above at pp. 205-07.) 

i 

2-: In December 1963 GE announced its 400 series. (DX 488; DX 490.) 

3 ;■ That series had evolved from work done in the Computer Department xn 

4^1 Phoenix in the early 1960s. (Weil, Tr. 7238-39.) The 400 series 

i 

5 ! was called the "GE line of the future which would be compatible 

! 
S j throughout". (PX 353, p» 44.) The 400 line was aimed at, among 

7 { others, IBM 1401 users. (Weil, Tr. 7031-35.)* According to Weil, 

i 
S i the GE 400 series (which was not compatible with the 200 series) 

9- | "was intended primarily for business data processing users, although 

10 : it did have some features that could support engineering and scien- 

XI | tific calculations, but strictly as a secondary objective". (Weil, 

12 i Tr. 7018, 7038.) 

13 I However, within a few years after the announcement of 

14.! IBM's System/360, "the distinction between a scientific computer and 

I 

12 \ a business computer . . . had been erased". (Weil, Tr. 7188-89.) 

! 

15 ■ GE was marketing the 400 for both scientific and business applica- 
17 ! tions : "Can scientists and businessmen be happy with the same 
^g || computer? Ask about a GE-400. Many installations have proved the 
^g ; GE-400 can handle engineering and scientific problems as easily as 
2q •! business problems." "So you see the GE-400 's don't just mean busi- 
2t || ness. They now offer you the broadest capabilities available today 

II 

22 S 

23 | * GE offered a "1401 simulator [with the 400 line] , a piece of 
j software which . . . had some hardware assistance which permitted 

24, \ programs from IBM 1401 [s] either to be run or to be converted easily 
1 to the 400". (Weil, Tr. 703.1-32. 



25 : 



) 



-491- 



1 
z 

3 
4 
5 

6 
7 
8 

ia 

12. i 
13 



on a medium scale information system — all the way from everyday 
business runs to complex scientific problems. " (DX 489.) The 
reason for this marketing change was, according to Weil, that " [a]s 
of 1967, the [IBM] 360 had been on the market for three years and 
the market in the middle range ... of computers was now much more 
homogenized between business and scientific than it had been earlier, 
and the GE 400 was hence sold as much as you could to a broad market 
encompassing the middle class of . . . engineering and commercial 
applications, both." (Weil, Tr. 7263.) " [S]o long as the scale of 
problem is suitable to the machine the machine could do either 
business or scientific work. The distinction between those two in 
this class of machine had largely been erased by that time." (Weil, 
Tr. 7264.) 



74 <[ • According to internal IBM reports , GE also reacted to 



IS '' 
16 
17 
13 

19 j 

20 : 



IBM's 360 announcement by reducing the price of the 400 CPUs between 
8% and 17% and the tape drives and their controllers between 14% and 
27%. (PX 2966, p. 3.)* The IBM Commercial Analysis Department 
reported that "[t]he price reduction gives the GE 400 a price/ 
performance advantage over comparable System/360 configurations. 
The improved price performance of the GE 400, coupled with 4-6 



21 i| 

;! * See also DX 1525, p. 1 (7/29/64): "GE has not officially 
*— ;j reduced prices, but they are selling their 400 line at 18% off. 

i They have also reduced their extra shift to a 10% charge"; and 
23 ;j PX 320, p. 16 (6/23/64): "The 400 line is a competitive offering 

.j today, but will require some revision if it is to remain competitive 
*"* ■ in the direct access market, and in the mixed business and scientific 



25 



environment of two years from now. " 



-492- 



L 
Z 
I 

•z 
T 
S 
9 

ia 

11 ! 

12 j 
13 
14 

IS 



months delivery, demonstrable hardware, and programming support 
makes the GE 400 extremely competitive with IBM's commercial product 
line*" ( Id.- ) This Commercial Analysis report evoked disagreement 
within IBM as to the effect of the GE price reduction. Khaplund 
"felt that, while in some applications the price reduction did 
indeed give the GE 100 [sic] a slight advantage, basically and 
broadly the reason for the competitive announcement was that our 360 
put them under pressure and they had to reduce the 400 for them to 
stay competitive"* As a result, -IBM' s President, A.- L. "Williamsv 
chided the President of the Data Processing Division, F. T* Cary, 
for disseminating reports that were "unduly negative". (PX 2966, 
pp. R-l, R-2.) 

GE initially announced four models in its 400 line of 
"compatibles"? in fact, however, only two were ever delivered. (PX 
353, p. 44.) Subsequent GE product announcements (the 600 and 100 
series) were not compatible with the 400 series. (Id.) In 1970, 



is . ;} 

'I GE cited the failure to deliver all the 400 models which had been 

17 ; 
is ! 



announced, as well as the incompatibility between 400 and 600 series 
computers, as yet another reason why GE developed an "image of failCing] 



13 ; 

I to follow through" in EDP. ( Id. , pp. 43-44.) 

20 '.\ — 

f 
I b. The GE 600 Series. GE announced its 600 series in 

21. ] - 

j July 1964, after the announcement of System/360. (Weil, Tr. 7197- 
22 j 

:| 98; DX 491, p. 1.) At that time the 600 Series consisted of the 

j GE 625 and 635, which differed only in memory speed. Later, GE 
[announced the 615, a "special configuration, slower memory speed 

?c 1 



-493- 



L 
Z 

3 

4* 

5 
6 

7 

a 

9- 
10 

li 

12 

13 
1* 
IS 

16 t 

17 1 
IS ! 

19 j 

20 ; 



version of the same 625/635 system", the 645, "associated with MIT 
in Project MAC", discussed below, and, eventually, the 655 which 
reimplemented. the 625/635 in higher speed integrated circuits. 
(Weil, Tr. 7198.) 

The GE 600 series also included "several compatible, but 
physically different military versions". (PX 4829, p. 18.) In a 
report on the 600 series, Withington wrote that the 600 series (and 
the 400) show "the same design emphasis on well-balanced, practical, 
but unspectacular systems. There are no technological innovations, 
and their basic speeds and specifications are no more than comparable 
to those of their competitors." ( Id. ) Internal IBM documents 
reported that GE was offering the 635 "at no extra shift charge". 
(DX 1525, p. 1.) When GE compared the 600 line against the announced 
IBM 360, it concluded that "depending upon exactly which model and 
details of usage and configuration, the 600 is either just a little 
more favorable or just a little less favorable than comparable 
members of the 360 series". (PX 320, p. 16.) As we have seen, 
however, IBM had made its own analyses of the competitive reactions 
to System/360 and improved its price/ performance with the 360/65 
before delivery. (See pp. 389-90 above.) 



2i j However, the 600 line was not as technically advanced as 



22 
23 

24 

25 



the System/360. Weil classified the 600 series as a "second generation 
solid-state computer". (Tr. 7192.) Moreover, in peripherals the 
600 series suffered in comparison to the IBM 360: 



-494- 



L 
Z 

z 

4. 
f : 

Si 
7\ 

i 

3- ! 

i 

9 i 

xa! 
u 

12 

13 

15 



"At present, GE's systems are somewhat handicapped because 
their peripheral equipment (particularly random-access file 
storage devices) is in some respects inferior to IBM's. GE 
says it is moving' actively to remedy this and to equal IBM's 
peripheral equipment with products of its own manufacture . " 
4829 r p." 1$.T 



Nevertheless, according to Weil, the initial customer acceptance of 
the 625 and 635 were "extremely good, well beyond our expectations"- 
(Tr. 7206.) 

One of the reasons for this was GE' s success with users of 
the IBM 7090/70 9 4 computers* GE had "carefully targeted as one of 
the markets for the GE 600 system the installed base of IBM's 7090 's 
and 7094' s" because the 7090/7094 "was at that time by far the 
leading scientific and engineering computer in the field, it had the 
largest number of such systems, so it was a large enough target". 
Further, since GE was itself a large user of the 7090/7094, the 
"members of these computer installations played a leading role among 
the user community of the 7090s and 7094s, so that ... we had an 
ia I; enormous resource to draw on who understood that market and the 
17 i needs of that user very well". (Weil, Tr. 7026-27.) 

;| 

•3 || GE "designed the 600 system to feel as familiar as possible 

15 ;j to a 7090 or 7094 user". Among other things, its peripheral equipment 

20 i could accept both media and format from such users and its software 

21 ! represented "a compatible superset, a software that would include 

•i 

22 j| the capabilities of what the user already had but would give him 

22 ii further extensions". (Weil, Tr. 7029.) To aid conversion, GE 
24. "t provided a piece of hardware "called a 7090 Simulator, so that a 
25 ! user who purchased this piece of hardware and put it in his system 



-495- 



I 

2 
3 

5 
S 

7 



9 ! 
i 

t 

u ! 



13 
14 
15 

is ; 

17 

IS 



could in fact run programs from the 7090 or 7094 without modification, 
or at least that was the hope. Most of the time it succeeded." 
(Weil, Tr. 7030.) 

As a consequence,* when IBM announced its 360 line as 
incompatible with its own earlier series, the computer group at GE 
was "initially at least overjoyed with what had occurred because it 
meant right at the time we were introducing a system designed to 



3 I displace 7090*s and 7094's, IBM had itself abandoned the 7094 and 7090 



computer series and brought out an entirely different computer 
series, and it was our belief at that time that it would be easier, 
if you were a user, to convert from the 7090/7094 to the 600 series 



1£ I than it would be to convert to IBM's new 360 series. We regarded 



that as a fortuitous occurrence and potentially to our advantage." 
(Weil, Tr. 7060-61.) The user of the 7094 was "forced ... to 
either go to a 360 or to some other competitive system, and we were 
sitting there with a system designed to make that conversion as easy 
as possible." (Weil, Tr. 7062.) That, of course, was one of the 
risks that IBM was taking with the 360, and by 1964 GE with its 400 



.Q '! and 600 and Honeywell with its 200 were attempting to take advantage 

2Q : of the 360 's incompatibility with previous IBM lines. 

2«f :j Weil testified that GE was "relatively successful in 

22 ;| converting user programs from the 7094 and 7090 to our 600 line" and 

2- |j that "our users found the conversion to involve work but to be 

within reasonable difficulty". "I don't really think we found a lot 
more difficulty . . . than we anticipated." (Tr. 7037-38.) Weil 



24 

S«5 



-496- 



L 

a! 

4* 

5 

i 

a 
ia 

11 ! 
12 

13 
14 
^ ! 

16 ! 

17 i 

IS I- 



estimated that GE acquired between 10% and 20% of the IBM 7090/94 
base. (Tr. 7269.) 

In addition to providing compatibility with the 7090/7094, 
the GE 600 (as had the 400) provided a compatibility feature which 
assisted conversion from the IBM 1401 to the 600 line. This would 
enable users who had previously used 1401s as off-line devices in 
conjunction with the 7090 or 7094 (e.g., tape to printer, peripheral 
operations) to move both the applications previously done on the 
7090/94 and the off-line functions run on the 1401 onto a single 
computer in the 600 line. (Weil, Tr. 7031, 7034-35*) 

The GE 600 series marketing strategy probably was based in 
part on the ability of the 7090 users who leased the 7090 to terminate 
their lease in a relatively short time and send the IBM equipment 
which they were using back to IBM. (Weil, Tr. 7207.) In Weil's 
judgment, "the GE 600 competed well with the IBM 7094". (Tr. 7212.) 

Naturally, in competing for conversion of the 7090/94 
customer as well as for other business, GE was competing against the ' 
newer System/360 IBM computers as well. Weil testified that "gen- 
^g ij erally we were competing with the upper end of the 360 spectrum as - 
2g ij it then existed. That would include occasionally the Model 50 but 
2i ij primarily the various models of the 60 's and occasionally the 70 *s 
22!; within the IBM 360 family". (Tr. 7207.) The restricted configuration 
23 i| 615 may also have competed with the 360/40. (Weil, Tr. 7209; see 



24 

25 



J also Tr. 7215.) Still latar, in about 1970, the 600 series competed 

I 

\ against the 370/145, 155 and (less frequently) the 165. (Weil, Tr. 



-497- 



I 
z 

z 

5 
S 
7 
S 
9- 
10 

11 s 

12 i 

13 
14- 
15 
16 



7210-11, 7215.) 

In "targeting" the 600 line against the 7090/94, GE in 
part paid a price for its success . Weil testified that the GE 600 
competed with the 360/65 "perhaps less well" than with the 7094 
"because it was very specifically targeted at the 7090/7094". 
(Tr. 7212.) 

GE described its 600 line as a "family of large-scale 
computers for business, scientific and real-time use" (DX 491, p. 1) 
and as "a new, advanced, high-performance, large-scale computer for 
use in business, scientific, and real-time applications — complete 
with all software". (DX 492B, p. 3.) Weil testified, however, that 
the line was originally intended "primarily for engineering and 
scientific computation, but with specific features that would make 
it attractive as well for business and commercial application, 
but that in this case was the secondary market". (Weil, Tr. 7019.) 
"While the machine was basically a scientific machine derived from 



.- i the 7090/7094 we were trying to replace, we also included extensive 



IS 

19 ! 



character manipulation facilities , which would have been typical of 
earlier business machines, commercially oriented machines, and was a 
- n part of the support we built into the machine — there were other such 
9T |j supports — part of the support we built into the machine for a COBOL 
compiler to make it attractive to business applications". (Weil, 
Tr. 7192.) GE "had the ability to use the growing low cost of logic 
•i to provide a number of features aimed at these several markets" 
;{ ( id. ) , reflecting the fact that "since the early sixties it really 



22 | 

23 i 
24 
25 



-498- 



L. 
Z 

Z 



hasn't been economically important to design a computer system only 
for business or only for scientific applications , except at the 
extreme ends of this spectrum, where you were trying to do as much 
scientific calculation as you possibly can within the limits of the 
technology". (Weil, Tr. 7190.) Thus, GE' s "intentions" with 
respect to the scientific marketing emphasis of the 600 series were 
T |j differences of degree, not of kind. The perception at the time was 
fl . j that the 600 series, like the 360, would compete in all application 
areas. 

Thus, Withington in 1964 wrote: "GE also believes (and we 
agree) that in^the large-computer area there are no longer significant 
distinctions between scientific and business machines, so the poten- 
tial market for the 600 series and its successors is very large". 
(PX 4829, pp. 18-19.) He also wrote: 



S- 
10 
II ! 
12 
13 
14- 
15 



"GE's product line, then, is more analagous [sic] to IBM's 
than that of any other competitor. GE hopes to compete not by 
being different, but by doing the same things better: by 
lo ;{ providing a combination of hardware, software , price , and 

I customer service which will appear superior. No competitor 
17 ; desiring a rapid increase in market share and profitability 

'{ could afford to follow this approach. However, GE has repeatedly 
IS ii stated that its intention is to build a solid and major position 

i in the computer industry: its approach is consonant with this 

IS ;! goal." (Idw p. 18.) 

'I 
2C J And Weil made clear that the 600 was marketed after its 

21 ■] announcement for both business and scientific applications: "[A]s 

22 "> the 600 was sold, as it went on in its lifetime, it was sold more 

1 

22 i and more to organizations that were more business-oriented and less 

t 

24, j scientific-oriented, partly as a result of bringing it down to the 

:j 
25 I 615 . . . which was more in the territory of more business installa- 

1 
i 

1 -499- 



L 
2 
2 

4 

5 
6- 
7 
3 
9- 

ia 

IX 

12 I 
13. j 
IA 
L5 



17 1 
19 il 



20;! 



tions, and partly because we found there were many more business 
customers, out there than were scientific customers for our class of 
system." "[Tjhe customer base that we built* up became more and more 
business-oriented with time,." (Weil, Tr.. 7270-72; see PX 328, 
p- 21.) 

Weil testified that the 600 also had capabilities for real 
time applications, which later turned out to be very useful for time 
sharing (discussed below) although those capabilities "were used by 
very few of the actual users that we sold the machine to". These 
real time capabilities were "a direct reflection of the military 
parentage of the central processor and the memory controller portion 
of the 600 system". (Tr. 7192-93.) In particular, the development 
of the 600 line drew on the work which had been done for the GE M- 
236 military computer by the Heavy Military Electronics Department 
in Syracuse.* (Weil, Tr. 7178-79, see Tr. 7301-02.) Among the real 



tg !| time uses of the GE 600 were the data reduction and monitoring done 



in connection with the Apollo launch system. (Weil, Tr. 7200; DX 
556, p. 5.) 

Even with the initial announcement of the 600 series, GE 
was thinking about the importance of time sharing as an emerging 

2-. :j area. In July of 1964, in an internal GE publication, the General 

ii 
22 !| 

2^ ; * Weil testified that "for the hardware aspects of the central 
; processing units", the componentry, skills and the manufacturing 
i facilities required today to produce the central processing unit are 
1 essentially the same whether one is speaking of a computer which is 
used for scientific, commercial or process control application. 
(Tr. 7191.) 

-500- 



25 



L ■ Manager of the Computer Department was touting the 600's ability to 

2 : permit "a large number of low-cost remote stations [to be] connected 

3 ; . . . by common carrier lines, thus permitting many people access to 
4- j the computer's problem solving skills. The need for many small 
5 it computers on college campuses , large government installations , or in 
a ■ widely-dispersed manufacturing organizations might thus be elimi- 
7 || nated." (DX 491, p. 1.) While "the system had a number of features 
3 1 in its peripherals and architecture which would make this possible 



9 

10 
11 
12 

13 



16 9 



... at the time of this announcement we [GE] did not supply a 
system that could support such an application". The hardware features 
included "an excellent form of memory protection to isolate the 
system software from whatever users may be doing and to provide 
memory relocation features so we could accommodate a number of 
different programs in the system at the same time". As we shall 
see, the 600 was subsequently used in this way — as a time-sharing 
system. (Weil, Tr. 7199-203.)* 

Notwithstanding the attractiveness of the 600 series 



17 s| 

•I conceptually, GE encountered difficulties in delivering the 625 and 

:t 635. Weil testified that: 






t * As Weil explained, at the time it was thought that a single, 

21 j central, shared computer was more efficient than a number of smaller, 
i stand-alone computers. (Tr. 7203-04, 7254.) Thus, the 625 and 635 

22 .; were "actively marketed for remote batch applications" as a cen- 
; tralized system in which it was contemplated that remote batch 

23 j terminals would replace earlier smaller stand-alone systems. (Weil, 
| Tr. 7252-54.) In the 1970s with lower and lower hardware costs, the 

2- ] trend turned the other way with many people believing that a number 

■| of smaller computers were more efficient than a single large computer. 
25] (See below atpp. 1276-86, 1339-40, 1448-59, 1510-16.) 



-501- 



"We were attempting to bring to market simultaneously a 
new central hardware system, a new processor system, a new set 
of peripherals, and an entire new set of software. 

"On top of that this was the first time that General 
Electric had ever attempted to put together and market so large 
a system, and as a result of all of those factors at once, we 
had a great deal of difficulty making the systems perform to 
our customer's and our own satisfaction in the field. A combined 
set of hardware difficulties and software difficulties", 



including 



" . . .a lot of difficulty with the magnetic tape units , we had 
some unreliability in the memories we were using", 

and, because of the size and complexity of the system, 



\ 

} §i 

"one of the difficulties we had was when something went wrong 
J ; we had the problem of telling just what had gone wrong in this 
roomful of equipment, so diagnosis was a problem for us as 
well." (Tr. 7215-16.) 

1 '. The difficulty with the software "centered around the operating 

2 j system called GECOS, which was . . a comprehensive operating 
4- supervisor", among the first of such systems. "[I]t was ambitious 

in its design. We had a great deal of difficulty in getting it 
5 • built, made reliable and made efficient." (Weil, Tr. 7216-17.) 
7 ;J There were three versions of GECOS. GECOS I, which had 

S I originally been intended for the 625 and 635, was never brought to 
a j the field. 

[Q j "It died in our test rooms because it was clear that it 

| was sufficiently scrambled up internally that it would not make 

rt .j a good product, and so GECOS II was constructed to take its 

.1 place using the lessons that we had learned on GECOS I. 

j "GECOS II was the first version of GECOS that was sent to 

the field, and while it had a good deal of difficulty when it 
went to the field, eventually, with much patching and baling 
7i I wire, was made to operate satisfactorily. 



LZ 



15 \ 



-502- 



1 

2 
3 

5 

6 

7 

a 

9- 

10 
11 



"GECOS III was initiated at that same — at the time period 
that GECOS II was in the field again to make use of the lessons 
we had in bringing GECOS II to the field, to reflect them back 
in what we hoped then would be a clean design and a clean 
product, so that GECOS III would incorporate the lessons of our 
field experience. 

"It was started and it was brought to the field much 
later, I believe around 1968 . . . . " (Weil, Tr. 7 217-18.) 

Weil echoed the theme of many computer people during the 1960s when 

he said that GE's problems resulted at least in part because it was 

attempting to develop a state-of-the-art software system. (Tr. 

7217-19; see Perlis, Tr. 2001-04; McCollister, Tr. 9694-97, 9706-08; 

Rooney, Tr. 12132-36, 12349-50, 12358; Conrad, Tr. 14088-89, 14133; 

Withington, Tr. 56727-30.) 



12 i The difficulties encountered with the 625 and the 635 did 

13 I not result in slippages in delivery dates although Weil testified 

14, ! ' that "perhaps they should have. The difficulties occurred much too 

i 

15 i often out in the customers' installations." (Tr. 7220-21.) 

i 
15 ' In late 1966 or early 1967, the 600 series systems were 

tj ij withdrawn from the market and "put into . . . hibernation" ("it was 

ia I put to sleep for the winter"). GE continued to support the systems 

;i 

^g : already sold but did not actively seek new sales. That winter sleep 

2q '; period lasted for at least a year or two, and the systems were not 

2i ij marketed again until 1968. (Weil, Tr. 7221-22.) 

22 I; l n the fast-moving computer business, withdrawal is a 

9«s '(mistake. Whereas IBM, when confronted with similar difficulties, 

•! 

24, I P ut a H °f its effort into solving them and keeping its customers 

.1 
3- 1 satisfied (see above at pp. 3 71-72 ) t GE withdrew. Weil said that 

j 

•i 

:! -503- 

.i 

i 
'i 



1| "the hibernation of the 600 was a mistake"; "it led to a considerable 

i 

2 | undermining in the confidence of General Electric' s offering of this 

i 

3 ; class of system" and adversely affected GE's image in the computer 

4-j industry. Weil explained why: 

i 

5 ]■ "When you buy a computer system, you are expecting a great 
deal from the man who — the company that supplies it to you. 

6 i You want to make sure that they will still be in business; that 
they will stand behind any difficulties that your system has 

7 j had, and that they will make it do what they told you it was 
going to do. And any indications that people were backing away 

8 | from such a full commitment would surely reduce a customer's 
confidence in that particular vendor." (Tr. 7224-25.)* . 

9 

10 i 
11 
12! 
13 

15 
15 
17 
13 
19 
20 
21 

22 'i ■ I 

\ ! 

23 I 
; * Withington, when he testified, emphasized that the customer's \ 

- A \ relationship with a computer systems vendor depended on the customer's | 

"* ■' understanding that the vendor was "credible" — that is, that "a given j 

manufacturer is a good one to be associated with . . . over time". i 

(Tr. 55735-36, 57671-72.) i 

i 

i 

-504- ! 



Withington also testified "that the inability of the 
General Electric Corporation to deliver operating systems [including 
GECOS and MULTICS] which completely met their advertised capabilities 
hurt the credibility of General Electric." (Tr. 56754, 56727-28.) 

There were direct financial consequences as well. In 1966 
GE reported that "in the information systems business, current 
operating losses were higher than projected because of difficulties 
involved in meeting a very sharp increase in shipments, and because 
of expenses in integrating worldwide product offerings. Substantially 
increased costs were also encountered in getting some new systems 
into operation." (PX 327, p. 9.) John L. Ingersoll, who had been 
Financial Manager of GE's information systems business in the late 
sixties, testified that from 196 5 to 1968 GE's difficulties with the 



25 



L 
2 
3 

4- 
5" 

3 :» 
7 

a 

9 
10 
II 
12 

LA ' 

15 
IS 
17 
IS 

19 



V) 



24. i 



625 and 635 "were a major element in the financial results experienced 
by that segment of GE". (Tr. 8339.) 

Such difficulties , experienced with the first computers 
and software of the 600 line, were aggravated when it came to the 
development of time sharing. 

c. Time Sharing. GE was involved in "two somewhat separate 
threads" in the development of time sharing. (Weil, Tr. 7106.) The 
first of these, developed by Dartmouth with some help from GE, was 
"a very effective small time sharing system which we then brought 
into our engineering organization and eventually modified, documented 
and offered as a product . . . initially on a system derived from 
the 225, later on a system derived from the 235, and eventually, 
very related, conceptually related systems were offered on the 400 
line and on the 600 line". Weil believed that this was the first 
commercial time -sharing offering. (Tr. 7106-07.) 

That system was "independent of the separate path which 
it involved the more ambitious, technically, time sharing system based 

;| upon the 645 and the MULTICS software". (Weil, Tr. 7106-07.) That 

it 

ij more ambitious development involved Bell Labs and M.I.T. (Weil, Tr. 

■i 

I 7108, 7225-26, 7231.) 

,i 

I Early in 1964, the Project MAC organization at M.I.T. , 

i 

I which had already developed a time-sharing system (CTSS) on a pair of 

I IBM 7090s (Brooks, Tr. 22739-40; Perlis , Tr. 1881, Weil, Tr. 7226-27), 

1 was "interested in developing an extremely advanced time sharing 

system". It approached a number of manufacturers "for a cooperative 



-505- 



L 

m 

3 
x 

7 i 

3 

.0 
.1 
2, 



effort in that development". (Weil, Tr. 7108.) General Electric 
proposed to Project MAC a version of the 635 system, which "would be 
modified in accordance with some of the discussions we had had with 
them, and which would provide then a hardware base for the advanced 
time sharing system they wished to develop . " In addition , GE 
"proposed working jointly with them in the development of the software 
that would reside on that hardware." (Weil, Tr. 7111-12.) 

In the summer of 1964, Project MAC selected GE over bids 
from IBM, DEC (who did place a $1 million peripheral processor, 
however), and others. (PX 2961, pp. 1, 3.) IBM believed that its 
rejection was. due, at least in part, to the fact that it had proposed 
to implement time sharing without dynamic relocation hardware. 
(Knaplund, Tr. 90533-35.) Weil confirmed that GE believed that 
"certain aspects of the 600 architesture [sic] , the 600 system, as 
laid out, were more amenable to some of the things that MIT wanted 



17 ; 

** il 

\ 
t 



,^ i to do than were either the 7094 based system or the 360 based 

system". "[W]e had a good meeting of minds, a good agreement on 

It philosophy with the Project MAC team." (Tr. 7115.) Project MAC and 

GE — and others in the industry — believed that computer systems were 

evolving toward "an information utility" based on the time-sharing 
20 ' 

i concept which would be of crucial importance to the future of com- 

i puting. (Weil, Tr. 7116, 7251-52, 7254-55; see Perils, Tr. 2117-18; 
22 "I 

: PX 320, pp. 9-10.) Wright, Director of Time Sharing Marketing for 

23 

\ IBM from 1964 to 1965, summarized his eight or twelve conversations 
24 I 

II with Dr. Ivan Sutherland, the Director of Information Processing 

?e 'i 

| 

■! -506- 



Techniques for the Department of Defense's Advanced Research Projects 

Agency (which funded Project MAC) (Tr. 13287-90) as follows: 

"Ivan Sutherland was essentially exploring what IBM was 
doing in the timesharing field. . . ' • I think that he was 
trying to convince himself whether or not we were . serious, 
whether or not we intended to follow through with a degree of 
urgency in the project. 

"He spoke words of encouragement, encouragement in the 
fact that he believed that IBM should pursue development of the 

time sharing concept in products and software as a matter of not 
only great importance to the United States government, but also 
of great importance to IBM and he simply encouraged and wanted 
to be kept aware, sort of as an insider, of how things were 
going on the project • 

"[I]t was my understanding that his interests were the 
fact that he believed providing time sharing facilities to the 
Department of Defense contractors in design" of new weapon 
systems, and use in other things, including health systems and 
so on, would, in fact, foster the use of computers, but, more 
i importantly from his standpoint, would assist in the solving of 
problems that these people in their research and development 
activity were confronted with and the use of computers would 
facilitate the solution of those problems at a more rapid rate 
and, therefore, accelerate the advancement of technology. 



1 ! 
i 

*! 

z i 

5 I 
* 

! 

Si 
I 

I 

7 1 

i 

3 | 

9- ! 

I 

ia ! 
ii 

12 

12 

IA 
15 
16 

17 

tQ !! 

19 | As shown above (at pp. 417-36), many people within IBM 

^ u ;| also believed that the time-sharing computer utility concept might 

<"• ;! well be the wave of the future and failure to respond to competitive 

;i 

*— ;j thrusts in this area — especially by a competitor with the power and 

--. it 

^ j potential of General Electric — might relegate IBM to a secondary 

24 j position in the future. Thus, for example, in September 1964, 
ij 
! responding to the loss of Project MAC and of other important accounts 

-507- 



"[I]t was clear that he felt that two large companies, 
such as GE and IBM, pursuing developments in time sharing, was 
beneficial to the government, was beneficial to industry and, 
therefore, that he thought that was a good situation." (Tr. 
13290-92.) 



25 



in the time-sharing area, Nat Rochester, a member of IBM's Time 
Sharing Task Force, concluded: "There is much more at stake than 
these few prestige accounts. What is at stake is essentially all 
computing business, scientific and commercial. . . ." (PX 1194A, pp 
2-3.) Two days later, the Research Group of the Time Sharing Task 
: ! Force wrote: 



"There is a very strong probability that the 'computing utility 1 
will be the way of all scientific computing in a few years, and 
a good possibility that it will capture a substantial part of 
the commercial market as well. IBM cannot afford to overlook a 
development of this scope. We are currently in danger of losing 
all contact with the leading developers of this concept. " (PX 
2811, p. R-3; emphasis in original.; 

Similar thoughts were expressed by other groups within IBM. The 

Scientific Computing Department reporting on "remote scientific 

computing" urged: 

"Certain accounts have already been lost. A small set of 
.4 '.[ key accounts are right now in the process of evaluations 

leading to computer acquisition decisions. For every such 
case, decisions disadvantageous to IBM appear to be in the 
offering. In quantity, such losses do not appear to be large. 
,3 . In quality, they will have a tremendous impact upon a very 
large market . segment. 

.7 ; 

.3 ■ 

;j "If we do not respond on the time-sharing requirements in 

^g ; the near future, the time-sharing market will be largely lost to 

I GE who has responded to this requirement. A large part of the 
balance of the remote scientific market will also be in 



sS 



•i 



j jeopardy." (PX 2964A, pp. 4-6.) 



■j Wright put it concisely: 

;j 'And all during this period of time, in general, the 

,- ;j industry was in a state of agitation because time sharing 
" - i appeared that it might indeed be a new wave of the future from 
^ '[ the standpoint of computing facilities for a company or an 
institution. 



-508- 



c- 



i ! 

i 

z\ 
z\ 

4-i 

i 

5" | 

t 

I 

5 ; 

7 I 
3 

NX 

11 
12 ! 
13 

14 
15 ! 

IS . 
17 ! 



"So there was clearly, you know, an understanding that if 
IBM for some reason did not respond to this particular require- 
ment of customers 1 need, demands of the customer, it was very 
likely that those customers might very well buy such capability 
from somebody else. 



"[T]he significance would be that IBM would lose business 
and that part of the installed base that IBM had at that point 
in time would disappear." (Wright, Tr. 12843-45.) 



In addition to being in the forefront of the new wave, GE 
expected two additional benefits from its work with Project MAC. 
First, "it was an opportunity for us to work with one of the organi- 
zations that was widely regarded as an advanced thought leader in 
the field, hence, we hoped to benefit technically from that work, 
but also because it was based upon 600 line hardware, even though it 
was largely incompatible with the 625/635, it would nonetheless 
provide a reflection on the 635 and 625 hardware in the minds of our 
prospective customers, so that the customers would feel that the 
machines they were buying were related to and that he might someday 
look forward to growing into the kind of applications that MIT and 
, g || GE were developing on the 645". Second, "it lent an aura of advance- 
ment to the rest of our commercial offerings." (Weil, Tr. 7122-23.) 

GE and M.I.T. were not the only participants in Project 

i 

[ MAC. Bell Labs was also to be involved in the development of the 



19 ; 

20 l! 

21 ; 
;! MULTICS system, "a system, hardware and software together, for 



22 ;| 

j| carrying out a very advanced form of time sharing, a multiple access 

.1 to extensive system facilities". (Weil, Tr. 7225-26, 7231.) 

The first GE system installed at M.I.T. was a GE 635, 



21 
25 



-509- 



L 
I 

I 

f 

3 

7 



which was "used as a development facility, but the project was aimed 
... at developing the MULTICS system, and a part of the MULTICS 
system was a special expanded version of the 635, which was later 
• termed the 645-" The 645 involved "major extensions to the central 
processor, primarily having to do with the way in which memory was 
addressed and accessed." "[T]here were hardware protection features" 
and a "high capacity input/output controller". "A very advanced 



2 \ form of dynamic relocation was included in the 645." (Weil, Tr. 



9 

.0 

2. 
3 

.4- 

'3 



7227-28.) 

In the fall of 1965, GE announced the 645 as a product at 
the Pall Joint Computer Conference. (Weil, Tr. 7233.) In December 
it announced that it was working toward the "broad commercial availa- 
bility" of the 645 system. (PX 326 (DX 13668, p. 15).)* However, within 
a year of the December 1965 announcement, the 645 was withdrawn "because 
we began to realize that what we had on our hands was a research 
project and not a product. . . . We were attempting to do something 



: that had never been done before, and, in principle, we might end up 

L7 ;, I 

j discovering that it was not feasible. As it turned out, it was hard | 

LS ;i I 

i and slow, but it was feasible." Weil described the GE 645 as "being j 

19 | i 

j in the research project stage" until 1969 or 1970. (Weil, Tr. j 

20 \ ' i 

I 7234.) j 

El ! ' 

'. I 

■J ! 

22 ;» ! 
:j * GE had already bid a version of what came to be the 645 to GM ; 

23 i Research along with time-sharing software, graphic console and devices, j 
j Neither the hardware nor the software existed at the time , and the ! 

24. i consoles were to be modified versions of those already being marketed 

! for military applications. (Hart, Tr. 80284-87.) ; 

25 ' 1 

I 

i ■ 

.! -510- ! 



L j 
i 

i 

i 

1 

3 ! 

i 

*•! 

i 

5 \ 

I 

6 I 

i 



3 

ia 

11 ! 

i 

12 j 

14 j 
i 

15 : 
| 

16 j 

17 j 

is ! 



At the time of the public announcement of the 645, the 
software had not been developed and the 645 itself was not in 
existence. General Electric never offered the 645 again as a product, 
(Weil, Tr. 7234-35.) 

While the 645 was intended "to provide a top-of- the- line 
prestige luster to the 600 line and to our other products, and also 



7 J to be a prototype for future sophisticated time sharing systems", as 



it turned out, "because of its lateness and its difficulty, it 
represented very little to General Electric except a drain on its 
resources" (Weil, Tr. 7236) although "some of the features that were 
pioneered in the 645 have since appeared elsewhere". (Weil, Tr. 
7237.) Despite the potential which it had for future success, GE 
never put its principal marketing thrust on the GE 645. (Weil, Tr. 
7236.) The 645 was never delivered, and Project MAC received a 
system designated a "636". (Wright, Tr. 13375-76.) 

Although Weil and others believed in 1964 and 1965 that 
the MULTICS system "could be technically feasibly designed", the 
participants in the Project MAC effort "underestimated the difficulty 
jg ij of successfully developing MULTICS". (Weil, Tr. 7232.) "[T]he 
2q i{ system operated in the way that [it] was originally intended about 
2t ,j three years behind its own schedule." Weil testified that this was 
22« a consequence, first of the difficulties of cooperation among M.I.T., 
2- j Bell Labs and GE, and, secondly, because "the technical task that 
« ij was being attempted was extremely sophisticated and many of the 



25 



subjects were at the state of the art as it was then known, and it 
took a long time to iron out the details of implementing some of 



■I -511- 



i; 

2 I 
3 

S 
7 
3 

10 

11 
12 
13 



these important features". (Weil, Tr. 7233.) Such problems occurred 
in other state-of-the-art software efforts, including those of IBM. 
(Perils, Tr. 2001-04; McCollister, Tr. 9694-97, 9706-08; Rooney, 
Tr. 12132-36, 12349-50, 12358; Conrad, Tr. 14088-89, 14133; Currie, 
Tr. 15704-06; Withington, Tr. 56727-30.) 

d. False Starts . On a number of other occasions during 
the 1960s, GE began development of product lines which were cancelled 
or greatly reduced. 

Weil testified that in the early 1960s, a series known as 
WXYZ was in development in Phoenix. "WXYZ was a series of four 
systems of which the Z was to be the most powerful." By the time 
Weil became familiar with it, "only the X and the Y were under 
serious development". After "considerable evolution", the X eventu- 



al ally became the GE 400. "The Y was to be a rather sophisticated, 



15 
IS ! 



larger system, but it was cancelled at the end of 1962 and its place 
in the market spectrum was eventually covered by the beginning of 
the 600 project."' Neither the W nor the Z was ever delivered. 
(Weil, Tr. 7238-39.) 



!j Following the announcement of the 600 series computers, GE 

_ '; considered a series of new product lines . An important event which 
20 : 

•j triggered these lines was the acquisition of overseas affiliates, 
'■! the Bull Company in France and the Olivetti Electronics Division 

22 ; 

(later known as GE Information Systems Italia) in Italy. (Weil, Tr. 
7239; see also PX 326 (DX 13668, pp. 3, 15); PX 328, p. 18.) 

According to Weil, GE at that time was interested in 



25 



-512- 



I 

2 

3 

4-i 

5 



producing "a world-wide product line which would cover the main 
portions of the product spectrum", and as a result a series of 
product lines were conceived. (Tr. 7239-40.) 

The first such line, the GE 100 line, was conceived during 
the tenure of Dr. Louis Rader. Rader joined GE in 1964 as Vice 
President and General Manager of the Industrial Electronics Division 



j \\ and took over the GE Information Systems Division which was formed in 



3 

S- 
10 

II 
12 



1965. (DX 13668, pp. 15, 33.) The GE 100 line consisted primarily of 
three sets of processors which were to be manufactured in Italy, 
France and the United States and which GE intended to market through- 
out the world. (Weil, Tr. 7240.) In 1966, Rader was transferred 
from the Information Systems Division to become General Manager of a 



„ ! | new division, the Industrial Process Control Division. (PX 327, 



i. pp. 9, 33.) Despite the fact that several study groups recommended 
proceeding with the 100 line, Hershner Cross, who took over from 
Rader as General Manager of the Information Systems Division in 1966 
"overruled all the study groups and decided that the 100 line would 
be abandoned". Cross did this "at the same time that he put the 600 



IS 
IS 

17 

13 

j into hibernation." (Weil, Tr. 7223-24, 7240-41; PX 327, pp. 9, 33.)* 

IS I 

•t 



20 1 

j * Weil testified that the GE Italian computer operation pursued 

21 j the 100 line after the decision was made not to proceed domestically. 
■j The lower members of the line manufactured in Italy had their names 

22 'j changed several times and were brought t6 market originally as the 
.1115 and later as "successive members of a moderately effective, low 

23 [priced business system". (Weil, Tr. 7241.) The GE Italian opera- 
tion pursued the 100 line despite Cross's edict because "they had a 

2i | strong general manager". (Weil, Tr. 7242.) 
.1 



-513- 



1 ; After cancellation of the 100 line, GE began to consider 

i 

2 J different new product lines. 

3. I "Upon cancellation of the 100 line, one of the measures 

that was taken was to initiate a study centered in France, but 

4 ! with worldwide participation, to spec out a more advanced line 
than the 100 line that would serve the same general purpose. 

5 : 

"This project, known as Project Charley, met in Paris for 

5 | a period of a number of months , but nothing broader came out of 

that beyond a book of proposed specifications 

7 ' 

"At that point there were some management and personnel 

g | changes in General Electric and it was about at this juncture 
that John Haanstra came to General Electric, and he initiated 

g. i the development of another line of computers, again to be 
worldwide and again to serve a broad spectrum. 

10 i 

11 

*2 i The ERW line began in late 196 6 and "lingered on for a while after 

that, but its principal effort was for eight or nine months, beginning 



"Eventually this line of machines was known as the ERW 
line . . . ." (Weil, Tr. 7242.) 



13 
14 
15 

16 . 

17 ! 

13 ii 
» 1 



in the fall of '66, into the spring of '67". After that, 

"John Haanstra 's responsibilities were changed and he was 
put in charge of the Phoenix operation. He lost his personal 
identification with this worldwide product line and instead 
became a champion of what was going on in Phoenix, which of 
course was very heavily the 600. 

"The ERW line was largely leaderless for a period of time 



:} "Then Dick Bloch came to General Electric and he instituted 

l| a line, I believe initially called the 700 line and eventually 



21 
22 
23 
24 



called APL, which was his conception of a worldwide, broad 
spectrum computer line." (Weil, Tr. 724 3.) 

This was in 1968. (R. Bloch, Tr. 7615-16, 7755.) Haanstra, who had 

been recruited from IBM to lead the GE computer operations in 196 8, 

was moved to Phoenix less than a year later, then was killed in a 

plane crash in 1969, and none of these projects ever resulted in 

-514- 



I 

2 

3 

4 

5 

6 

7 

3 

9- 
10 
11 
12 
13 
14 
15 
IS . 
17 ! 



delivered products. (Weil, Tr. 7242-46; R. Bloch, Tr. 7756.) 

Bloch, who came to General Electric from Honeywell via 

Auerbach Corporation in 1968 and succeeded Haanstra as General 

Manager of the Advanced Development and Resources Planning Division, 

testified that when he arrived, there had been "several starts in 

the direction of an advanced product line". (Tr. 7592, 7611, 7757; 

PX 327, p. 28.) According to Bloch, while there were "some very, 

very excellent developments afoot" , the operation was poorly organized 

and "one would have wondered how this would ever be put together 

into a line". (Tr. 7757-58.) One problem was that there were 

various development activities under way under different auspices 

throughout the company. (R. Bloch, Tr. 7759.) 

"GE previously was typified to me as a company of great potential 
in terms of spot accomplishments in various areas — software, 
hardware, new attacks, in concept and in hardware too. But the 
real question was, how was it all going to be put together? 
That was one side of it. The other side of it was that nobody 
thought about the total plan, the total objective, what this 
business data processing world was all about." (R. Bloch, Tr. 
7759.) 



Thus, Bloch believed "the decentralized organizational 

^ !» approach of General Electric adversely affected their attempt to 

10 J 

l * || develop an integrated line of computer products" , and individual 

20 ; j departments took over responsibilities for obsolescent lines, promoting 

21 ;j their own interests. (Tr. 7759-60.) The problem was that GE, 



22 
23 

24 
25 



unlike IBM, was unable to tie together under central control this 
disparate collection of products produced and marketed throughout 



-515- 



the world.* What Bloch called GE ' s "decentralized organizational 
approach" was a substantial part of its downfall. 

e • The, Management of GE ' s Computer Operation . General 
Electric encountered substantial difficulties in managing its computer 
; operation. It had a revolving door of management personnel running 
its computer business during the period 1964-1970. During that time 
frame GE ran through a progression of managers and other key per- 
sonnel whose jobs constantly changed and who were succeeded by 
people with little computer experience.** The result was that 



^ i 



f I 

I 
I \ 

|| * The following statement from the 1968 GE Annual Report summarized 
the GE computer equipment business as of that date. 

7 { "On a world basis, General Electric offers five product 

lines, starting with the small-scale GE-50 series, produced by 
Bull-General Electric in France. The Italian operation produces 
the 100 series, and in 1968 introduced a more powerful system, 
the GE-130. The US-designed 200 series continued to hold wide 
acceptance for its dependable computing power. The GE-400 line 
of medium-scale systems is produced in the U.S. and France and 
in Japan under a licensing agreement." (PX328, p. 18.) 

'.[ ** In July 1963,. the Computer Department was headed by Harrison 
7 i Van Aken, who reported to the General Manager of the Industrial 

I Electronics Division. (Weil, Tr. 7085; PX 320, pp. 1, 2, 19; DX 485; 
3 ;jDX 491.) In 1964, Dr. Louis T. Rader was hired to be in charge of 

ijGE's worldwide computer activities. He was named Vice President and 
g ,, General Manager of the Industrial Electronics Divison and reported 

:|to Hershner Cross, Vice President and Group Executive of the Indus- 
g j trial Group. (J. Jones, Tr. 79357-58; DX 13667, p. 31.) In 

41965, General Electric realigned its organization and formed an 
£ l| Industrial and Information Group headed by Cross. Within that group 
~ j w as the Information Systems Division, headed by Rader. (dx 13668, 
"? :',PP» i ^' 33.) i n 1966, still another new division was formed, called 
— jthe Industrial Process Control Division, and Rader was transferred 
~ 'from the Information Systems Division to the new division to become 

tits General Manager. Cross remained group executive of the Industrial 
M ; |and Information Group and at the same time served as Acting General 
"~ {Manager of the Information Systems Division. (Weil, Tr. 7223-24; 
j- ;PX 327, pp. 9, 33.) Early in 1967, J. Stanford Smith, formerly Vice 

-516- 



2 ! 



3 ! 



projects begun were abandoned and no continuity of purpose or product 
development existed. But the problems went deeper than that. 

According to Weil, one of the "major mistakes" which GE 
• made in managing its computer business, stemmed from GE's "very 
strong" belief 



7'i 

3 



9 
10 

II 



"in the philosophy of professional management. This basically 
is that management is a profession and a good manager can 
manage any kind of business . 



"This in fact works quite well for a mature or gradually 
declining business, where a man put into a business can model 
his behavior upon that of his predecessor's and then make 
adjustments as he learns what's really going on. In a rapidly 
evolving business, however, his predecessor's behavior, especi- 
ally if it was unsuccessful, is a very poor model. And since 
he knows nothing about the business , he is a professional 
manager and came from Toasters or Welding, or whatever it may 
. be, elsewhere in the General Electric Company, he really could 
•^ If: not understand what he was managing. 



13 



-*" ij President of Marketing and Public Relations Services , became the 

General Manager of the Information Systems Division. (DX 13668, p. 33; 
PX 327, p. 33.) In January 1968, GE again changed its organization, 
going from 5 groups to 10 groups and from 29 divisions to 46 divisions. 
^ :! Hershner Cross's Industrial and Information Group was split into two 

jj groups. Cross remained Vice President and Group Executive, heading 
17 ';{ up the Industrial Group which included the Industrial Process Control 

;| Division led by Rader. Smith was promoted to Vice President and 
IS -[Group Executive in charge of the Information Systems Group, and John 

j Haanstra, who had recently come from IBM, became General Manager of 
13 -I the Advanced Development and Resources Planning Division within that 

Igroup. (R. Bloch, Tr. 7755-56; DX 13669, pp. 27-29.) In 1968, 
2£ | Haanstra became General Manager of the Information Systems Equipment 

: l Division (PX 328, p. 28) and Richard M. Bloch replaced Haanstra as 
21 [General Manager of the Advanced Development and Resources Planning 

.| Division (later the Advanced Systems Division) . (R. Bloch, Tr. 
22. :! 7623-25; PX 328, p. 28; DX 556, p. 30.) In early 1969, Hilliard W. 

J Paige, who was Vice President and Group Executive of the Aerospace 
23 [Group, replaced Smith as head of the Information Systems Group. (PX 

1 328, p. 27; DX 556, p. 29.) Later that year, J. F. Burlingame 
24. 1 succeeded Haanstra, who was killed in a plane crash, as Vice President 

jof the Information Systems Equipment Division. (DX 556, p. 30.) 



-517- 



L ; 

i 

zi 
i 

3" ! 

i 

I 

s i 

i 

7; 

i 

3 I 

9-1 
10 | 

11 ' 

12 

13 



15 
16 x 

17 ! 

i 

13 ; 

19 I 

20 : 

21 ; 

22 ; 

23 ! 
2± ! 



"But if you have a series of these managers above each 
other, they feel they are in trouble, they now must do some- 
thing. What can they do? They do not understand the business 
well. So the only thing they can do is to replace the man 
working for them. 

"So the net result of this was, as we got into difficulties-, 
especially in bringing the 600 to market thereafter, we had a 
sequence of people running General Electric 's computer business, 
none of whom, except when we come to Dick Bloch and John 
Haanstra — and, again, they were not in charge of the computer 
business but were key people — none of whom were experts in the 
computer business. Furthermore, we had a new one every eighteen 
months or so. 

"So that General Electric never developed experienced 
management that understood the computer business, and I believe 
this was a major part of why General Electric never learned how 
to manage the business properly." (Tr. 7247-49.) 

This philosophy led. to General Electric having "a great 

deal of difficulty ... in entering dramatically new fields", 



1 a i although it was "extremely successful in managing mature businesses 



and declining business". (Weil, Tr. 7259.) This is undoubtedly 
! associated with GE's decision, discussed below, to remain with its 
"core" businesses rather than continuing in computers. 

GE's management problems were perceived outside of GE as 
well. Withington testified: 



"I recall that General Electric, consistent with its policy of 
rotating managers between divisions , changed the senior manage- 
ment of its computer systems business at intervals of approxi- 
mately three years and I recall feeling that this was a poor 
practice as the incoming managers rarely understood much about 
the business at the time they would take it over." (Tr. 
56731.) 

John Jones, of Southern Railway, testified that although 

Southern Railway was a "very large customer of General Electric" in 

-518- 



1 1 

i 

*i 

i 

4-1 



5 I' ment that I was in at the Southern Railway Company and with the 



s-i 

i 

t\ 

s ! 

i 

11! 

I 

i 

12 

13 
14 
15 
16 
17 
13 
19 
20 



other areas, Southern Railway "did not seriously consider their 
computer equipment". (Tr. 79352.) In the middle to late 1960s, 
Jones reached the conclusion "that General Electric was not a viable 
competitor, not one that I would consider selecting in the environ- 



project that I had before me to complete". (Tr. 79353-55.) He 

testified: 

"[Tjhrough personal knowledge at several levels in the 
company, I was at least to some extent aware of the activities 
of the Computer Division of General Electric, and it was my 
view that there were some serious problems in terms of how they 
were managing that function, and it was my concern that I would 
not be able to obtain the support and continuing responsiveness 
from General Electric that I would judge to be critical in the 
system that we were considering installing. 

j "As a result of those concerns, despite the fact that we 

had been a large customer of General Electric in other areas, 
it was my conclusion that I did not want to take on the risk 
of, or what I perceived to be a risk, of considering installing 
General Electric equipment." (Tr. 79354-55.) 



Jones' views crystallized in middle to late 1967 ( id. ) , about the 
same time that General Electric' s difficulties were being made 
public in the form of the "hibernation" of the 600 system.* (Weil, 
Tr. 7221-22.) 
j In 1970, GE ' s future product plans (then known as APL) 

j recognized that among the "negative factors" which affected GE * s 

ii 

ji image in the computer industry were (a) GE * s management indecision 
\ and replacement", (b) GE ' s "professional manager" image, (c) GE's 

23 

24 '-l 
25 



! * Jones based his views on his personal contact with General 
Electric at the time. (J. Jones, Tr. 79355-60.) 



-519- 



l\ "lack of long-term commitment" and (d) GE's "loss of key personnel". 
Z | (PX 353, p. 45.) 

3; I Rotation of management meant a lack of continuity in decision 

4| making over time, but another difficulty, as already observed, was 
5 | decentralization of decision-making responsibility at the same point 
in time. Bloch testified that 



6 

7[1 
3 

9- 

ia 

ii 

12 1 
13 
14 
15 

IS s 

I! 

17 l 

IS 

i 

19 j 

20 ; 



"GE operated in a decentralized fashion, with profit 
centers usually at the departmental level, and for reasons which 
I do not pretend to comprehend, the top management of the 
company allowed these growths to occur of quite competent, in 
their own right, groups, both here and overseas. 

"Overseas, of course, one can understand some of that, 
because there was outright acquisition. But even here there 
were a multiplicity of centers and there was a proliferation of 
activity; multiple peripheral devices of the same general 
character being developed at different places at the same time; 
a lack of coordination from any central area whatsoever. 

"Our plan was, indeed, to make use of the facilities 
worldwide but to have it completely controlled and specified, 
all standards set, from the central operation in New York. And 
this was a new philosophy to them entirely. And if this was 
indeed a new philosophy to them, then I can understand why they 
had problems earlier." (Tr. 7646-47.) 



Bloch had "no question" in his mind that "the decentralized organi- 
zational approach of General Electric adversely affected their 
attempt to develop an integrated line of computer products". He 
encountered "substantial resistance" to his attempt to limit the 

:i decentralization. (R. Bloch, Tr. 7759-61.) GE's decentralization 
21 || . 

ji of responsibility within computers reflected its general management 

j| philosophy. GE's 1968 Annual Report stated that "General Electric 
23 > 

s had 170 decentralized operating departments focusing on separate 
:! aspects of world markets in 1968. Its production ranged over seme 

25 ; 

-520- 



1 

2 



3" j. * The Computer Department was always buried deep in the 



4~ 

3 
6:1 



3 

9- 
10 

11 

12 ! 

i 

13. I 
14 
15 
15 , 
17 I 



3,000 different categories of products and 200,000 different models 
and sizes." (PX 328, p. 7.) 



organizational structure. Back in 1963, the Computer Department had 
been within the Industrial Electronics Division which in turn, was 
part of the Industrial Group. Weil testified that there were "some- 



7 I thing approaching a hundred" departments in GE at that time. (Tr, 



7153-54; DX 485.) In 1968, GE formed the Information Systems Group, 

one of ten groups containing 50 to 60 divisions and, in turn, 130 or 

140 separate departments. (R. Jones, Tr. 8794-95.) Because computers 

had been so far down in its organizational structure and because it 

had so many other products to attend to, GE failed to mobilize its 

resources in computers to the extent necessary. Weil testified that 

among the "major mistakes which GE made in the management of its 

computer business" were two which related to this. First, 

"a lot of ambitious and difficult tasks were attempted which 
turned out to be more difficult and more ambitious perhaps than 
was appreciated when we started. 



Tg .! "Secondly, General Electric was never fully committed to 

!{ its computer business. It was always a business . . . that 

General Electric could live without. So that if troubles came 
or budgets were suddenly bigger than had been expected, there 
was always this reconsideration of 'Is this really a business 
} we want to be in? And how do we prevent this from draining the 
j profits of our other businesses? 1 It was not the strong commit- 
i ment felt by those of us actually in the computer business of 

^ ■■! General Electric." (Tr. 7247-48.) 

22 ; * 

mmm i| He testified that there were differences, for example, between GE ' s 
2^ ! ' 

:| 

:i commitment to the computer industry and its commitment to the atomic 

! 

power business to which it "manifested a greater commitment to 



19 j 
21 



25 ; 



-521- 



success". Nuclear power was regarded as "an adjunct to that core of 
business of the company" consisting of the supplying of power genera- 
tion equipment. "It was clear that the mission of the nuclear power 
business was: We don't know whether there is a business, but if 
r there will be a nuclear power business, you will be one of the 
leading competitors." On the other hand, the "equivalent charge" 
for the computer business would have been, "We are sure there will 
be a computer business, now you must demonstrate that you can compete." 
(Weil, Tr. 7174-76.) 

Similarly, Bloch testified that, when he joined GE in 
1968: 

"They were in the business. They had been in for some 
period of time furnishing general purpose computing equipment. 
My feeling was, however, that it was always tainted with some 
tentativeness or speculativeness on the part of the company as 
a long term commitment to the field. My feeling was that if it 
turned out to be a great success, the company would be delighted; 
if it turned out not to be a great success , the company could 
extinguish parts or all of its activity in the field without 
necessarily any great remorse." (Tr. 7623-24, 7616.) 

! | Reginald H . Jones , Chairman of the Board and Chief Executive * 

■'} 

'I 

! Officer of GE at the time of his testimony and a member of the 

'J 

Ventures Task Force which recommended GE's exit, as described below, 

i [ 

j testified that he (and his predecessor, Fred J. Borch*) had agreed 
i with their predecessor, GE's former Chairman, Ralph Cordiner, who 
\ said about GE's computer business: "General Electric 's mistake was 



3 ; 



| * Fred J. Borch was Jones' predecessor (R. Jones, Tr. 8752), not 
i I his "successor" as Jones mistakenly says at Tr. 8870. 



-522- 



L 
2 

3 

a 
7 
3 

1C 

£1 

12 
12 

12 

15 
17 
13 



19 



that it failed to realize the opportunity and therefore made an 
inadequate allocation of resources, both human and physical, to the 
business". (Tr. 8869-70, 8731, 87520 

Jones testified that "as early as the 1950 's, if we had 
increased substantially the technical manpower assigned to the busi- 
ness , if we had increased at that time the financial resources 
required for the business , they would have been much smaller in 
terms of absolute numbers than they would have been, let's say, some 
fifteen years later." (Tr. 8875.) Jones said: "We never did make 
the allocation of resources to the business that were warranted. " 
(Tr. 8874.) The contrast with both IBM's commitment to the business 
in the 1950s and its investment and risk-taking with System/360 in 
the 1960s is striking. 

f. GE's Position in the Late Sixties. In 1964, 
General Electric obtained approximately half of Compagnie Bull 
General Electric and Societe Industrielle Bull General Electric 
for $43 million. • (DX 13667, p. 16.) By the time of the Honeywell 
merger, revenue of the Bull companies was $206 million. 
(DX 554, p. 10.) In 1965, General Electric acquired the majority 



r interest in Olivetti -General Electric for about $12 million. This i 

! 



23., 

i subsidiary was formed from the electronic data processing business 

21 ■! 

\ and the electronics laboratory of Olivetti of Italy. (DX 13668, p. 3.) 

22 ;; 

i In 1968, General Electric changed the name of Olivetti-General 

^ ; 

! Electric to General Electric Information Systems Italia when it 

I secured full ownership of the Italian-based computer affiliate . By 

i 

'I -523- 



L 
2 

3 

4 

5 

ei 

i 

7| 

i 

3 ! 

i 

i 

9-1 

10 ! 

i 

i 

11 * 
12 

13 

! 

I*! 
i 
i 

i 

i6 i 

17 
IS 

is ; 

10 



1968, General Electric had research, engineering and manufacturing 

facilities at 13 locations in five countries with a worldwide sales 

and service organization of 8,0-00 employees. (PX 328, p. 18.) As- 

has been noted, the Bull subsidiary produced the GE-50 series and 

the 400 series*" and the Italian subsidiary produced the 100 series 

and the 130. (PX 328, p. 18.)' In 1968, General Electric also 

broadened its line of input/output and storage devices and extended 

its time-sharing services. By the end of that year, more than 50 GE 

time-sharing systems were in place serving about 100,000 customers 

in 17 countries around the world. GE reported that this area of the 

business was "growing even faster than the computer equipment sector". 

(PX 328, pp. 18, 21.) 

GE also reported that "the company's investments in computer 

technology have given us an expanded worldwide base in what has been 

characterized as the world's fastest growing business. Again, our 

developing capability to serve this industry is leading to further 

j| new opportunities." ( Id. , pp. 3-4.) 

ji 

■ GE's 1968 sales of information systems were "well above 

those of 1967 and with operating losses substantially reduced" . 



( Id. , p. 18.) For the year ended December 31, 196 9, the General 

I 

i Electric computer operations which Honeywell acquired showed a 



21 :i 

;| profit. (Ingersoll, Tr. 8329-30; DX 554, p. 9.) Those operations 



23 



25 



i| continued to show a profit for Honeywell in 1970. (DX 148, p. 1; 



* The 400 was also produced in Japan and the United States. (PX 
328, p. 18.) 



-524- 



1 1 



3; J member of the large-scale GE-600 line" which had "had its best year 



i 

si 

» 

i 
& I 

t\ 

i 

3 ! 

i 

i 

ii 

12 



DX 13977. ) 

In 1969, GE announced the GE-655, "the most powerful 



in shipments and orders" . According to the GE 1969 Annual Report, 
the GE-400 line also had a successful year. (DX 556, p. 13.) 

Despite these improvements, GE was still in trouble. Yet, 
if GE "did not appreciate the problem that was building in the late 
Sixties" (R. Jones, Tr. 8876) , others did. Withington wrote in 
1969: 



"During 1968, General Electric was able to demonstrate completely 
successful operation of its GECOS-III operating system for the 
625 and 635 computers. . . . The 625 and 635 (recently joined 
by a smaller 615) are continuing to sell largely because of the 
success of GECOS-III, but the machines themselves are obsolescent 
from the point of view of cost-effectiveness. It is to be 

£3 j| presumed that General Electric has in development compatible 

successor machines which can capitalize on GECOS-III, but which 

2* | will show better performance. When this new line is announced, 
General Electric will be in a position to make a strong resur- 
gence in the large machine area." (PX 4834, p. 29.) 



15 

IT \ 

i 

IS I 

H 

20 r 



Withington judged the GE-400 line "obsolescent" as well and said, 
"General Electric' s future position is dependent on the timing and 
success of the new line." ( Id. ) As we have seen, GE had made 
several false starts to the development of "compatible successor 
machines" and was not, in fact, "in a position to make a strong 



i 

;j resurgence" . 

|! As of 1969, GE had several incompatible lines, which had 



i| been "developed at different times in different places, and to a 

I 

,| great extent under different management". (R. Bloch, Tr. 7787-88; 

24 ;t 

_ ; PX 3 28, p. 18.) Bloch, who came to GE in 196 8, concluded that the 

25 1 

-525- 



L 
2 

3 
4 

f 
6 
7 
3 

ii 

12 

i* 

15 
15 
17 
IS 



GE 100, 200 and 400 series computers 

"were beyond their useful time in terms of state of the art. 
They were in place doing their work, except that we were simply 
facing the natural problem of the field, and that is with time. 
You get to a point in which the price/performance is so improved 
over equipment of days of yore that it is clear that those 
users are going to move to new equipment, and either you are 
going to provide that new equipment or your competitors are 
going to provide it." (Tr. 7761-62.) 

As Jones put it: "You had to bring out something that would exceed 
the price/performance of the existing competition because you knew 
full well that they were going to be moving ahead of you. It -is a 
constant leap frogging game." (Tr. 8866-67.) 

Bloch testified that although the larger 600 series had 
come out more recently than the other lines, the importance of 
smaller systems "far outweighed the significance of that 600 series, 
looking toward the future". This was because Bloch foresaw a tendency 
toward increasing decentralization and smaller processors, "which 
are smaller physically, they are smaller dollarwise, but they cer- 
tainly aren't smaller in terms of power when contrasted with the 
earlier days". (R. Bloch, Tr. 7762-67.) Thus, while the 600 "would 
!| be more appropriate for extremely large, powerful systems that are 



19 i 

! meant to be operated on a highly centralized basis", Bloch felt that 



20 •. 

'i GE needed to pay more attention to the smaller members of the line 

| CR. Bloch, Tr. 7768; see Weil, Tr. 7252-54.) 



22 

24 
25 



The Ventures Task Force, organized in late 196 9 to consider 
GE's future in the computer business, reported in April 1970 that 
"most current product lines are obsolete" and that GE had a "lagging 
technical position in mainframes, peripherals and manufacturing 

-526- 



Li 

j 

i 
i 

t 

i 
5 ' 

S- 

7 

8 . ! 

9- 
10 
11 
12 
13 
14 
15 
16 



™, 



19 i 

il 

20 ;! 



2Z ! 



process technology". (PX 331A, p. 18.) 

This was not a secret held within GE. GE's reputation in 

the computer industry had suffered badly from its management failures 

and product obsolescence. Thus, the Advanced Product Line Master Plan 

in January 1970 recognized that GE's image in the computer industry 

was poor: 

"General Electric has the reputation of the ' sleeping 
giant 1 of the information systems industry, with vast capa- 
bilities and resources which have yet to be marshalled for a 
determined attack on IBM. 

"GE's image is one of failure to follow through, as charac- 
terized by: 

" . An enviable image in the banking industry was 
built through the success of the ERMA project and GE's 
leadership in development of Magnetic Ink Character 
Recognition standards. This image was subsequently lost 
due to neglect. 

" . In 1963, GE assumed a leadership position in the 
area of communication systems and communications control 
concepts with the announcement of the DATANET-30. Subse- 
quently, GE has lost its leadership in the field by not 
following up with any improvements until the recent announce- 
ments in 1969. 



**' || " . In the area of system capability, GE coined the 

phrase, 'The Compatibles' . When the GE-400 line was 
introduced, it was characterized as the GE line of the 
future which would be compatible throughout. Although GE 
announced four members (GE-425, GE-43 5, GE-455, GE-465) of 
this line, it. delivered only two. 



:| " . Since announcement of the GE-400 line, GE has 

*• i{ made two other major line announcements: the GE-600 line, 

which is not compatible with the GE-400 line; and the GE- 
100 line, which is compatible , with neither. In fact, GE 
currently supports seven mutually incompatible product 
lines . 



23 

24- " . In 1964, GE recognized the way of the future by 

an aggressive advertising and promotional campaign with 
2- ; j regard to direct access. It indicated that direct 

il -527- 



F i 
i 

T\ 
3 

? 

J; 

i 

1 ! 

z 

4. 



access was the way of the future and announced a line of disc 
storage devices to support this assertion. Since then GE has 
not followed through on this commitment even though the initial 
prognostications proved to be accurate. 

"A brief summary of GE ' s image with respect to the various 
product lines includes: * 

"'GJyflMA 10 — an ideal model for a beginner. 

"GE-50 — excellent for new users, but no compatible 
upgrade • 

"GAMMA 30 — an obsolete machine with no compatible 
upgrade . 

"GE-200 — an obsolete line with no compatible upgrade- 

"GE-100 — a good family of products. 

"GE-400 — a relatively obsolete line with no compatible 
upgrade . 

"GE-600— a reasonably good line with a need for a 
higher member (a la the GE-655) . Good operating system 
software — among the best in the industry. 

"As long as the user is able to remain within a given one 
of the seven product lines, he is reasonably satisfied. 



"Measures of customer loyalty appear to fluctuate from 
,6 jj year to year, but are generally below IBM and appear to be 

•j consistently below the industry average. This loyalty is 
j ; understandably low when customers must move up from the product 

i line which they are currently utilizing." (PX 353, pp. 43-44, 
a ;l footnote omitted.) 

( g ; ! Unlike IBM which had integrated its product line in 1964 

?g jwith a single compatible line conceived, developed and marketed on a 
•i 

n .jworldwide basis, GE in 1969 still had several incompatible lines 
w jwith fragmented development and inadequate worldwide coordination. 

I 

2 « 9' The Advanced Product Line (APL) . Plainly, if GE was 

f 
74 jto overcome its problems, it needed a new product line. This was to 

?B ;be 31och's task when he joined the company in 1968. (R. Bloch, Tr. 
i 

;| -528- 



L 
2 

3 

5 
S 

71 

8 j 

i 

9- I 

I 

10 i 

i 

i 

11 ! 

i 

12. j 

13 I 
14 
15 
IS t 
17 j 
IS j 

i 

20 ; ! 



7757.) 

This new line, "initially called the 700 line and eventu- 
ally called APL . . . was his conception of a worldwide, broad 
spectrum computer line". (Weil, Tr. 7243.) It was to be "a single 
integrated line to be marketed on a worldwide basis". (R. Bloch, 
Tr. 7798; Ingersoll, Tr. 8104.) GE's plan was to achieve the "number 
two position" in the field: 

"We could not also see a company such as GE being satisfied 
with a. $50,000,000 business, say, in some convenient corner of 
the field, even if it were able to make a. profit there, which 
might indeed happen, because a business that size is insignifi- 
cant in the GE scheme of things." (R. Bloch, Tr. 7648-49, 
7799.) 

As a result of this goal, GE's APL was to be "an attack" "across the 

board". The new line was to "attack" everything from the $500-a- 

month rental to the $70,000- or $80,000-a-month rental, which, as 

Bloch put it, "is a tremendous range", "well over 90 percent of the 

total range". Of course, GE was attacking IBM and "in particular 

attacking the IBM 360 series, and not only the 360 series, but what 

we surmised was coming soon, and which became the 370 series." (R. 

Bloch, Tr. 7647-48.) Bloch testified that, had the APL line ever 

been completed, it would have been a "more ambitious ... or broader, 

more comprehensive, line than any that was in existence in the year 



21 ;! 1970 — or '69 . . . with the exception of IBM". (Tr. 7803.) 

*— ; Bloch felt a sense of urgency. about this mission and sent 

" j a telegram to various parts of the GE computer organization (including j 

*■ ■ ! riaanstra in Phoenix and Weil m Bridgeport) creating a special task j 

1 ' 

! force. (DX 540.) Bloch testified that he fait the situation was ■ 

' i 

\ ' ! 

-529- 

1 i 

I ! 

I ! 

I ■ ! 

■i . ( 



25 



1 i 

i 

2i 

i 

i 

! 

Z ; 

i 

4-1 

si 

i 

I 
Si 

1 

i 

7 

3 

9- 

10 

11 ! 

12 j 

13 

14 

15 ! 

16 

17 

IS '1 - . 

j of anticipated worldwide shipments would be made to users of earlier 

19 '•! 

!| GE lines, with another 35% to be shipped to users of competitive 

;i systems, chiefly IBM. The remainder were to go to new users. (PX 

* I 353, pp. 53-54, 57.) To effect the necessary conversions, GE planned 

\ to offer various emulation and conversion aids. (R. Bloch, Tr. 

~~ !l 

** 'j 7881-84; PX 353, pp. 53, 62-63, 67, 118, 119, 164-65, 171, 175, 178, 

24 j 179.) 

; i In order to induce IBM users to convert, the APL line had 

;i 

i 

!j 

! -530- 



urgent because 

"the company was on a timetable if it was to enter the field in 
a fashion which I thought was necessary, which meant that we 
had to fix the specifications, characteristics, and get the 
assignments made, development of assignments and so on, through- 
out the world, certainly before the end of that calendar year. 
« •- • 

"The importance was simply that of time costing the company 
its future position in the field. By delaying the time at 
which we could announce and ship these systems, we would be, it 
was my feeling and generally agreed, losing some of our current 
base. 

" Secondly r IBM was, I thought, much more vulnerable at the 
earlier time within this period, that is, in the earlier seventies 
and that every month that could be compressed with respect to 
the schedule meant an ability to tackle IBM more readily and to 
preserve our customers . . . the present GE CPL [current product 
line] customers who had obsoleting equipment. And there was 
the danger, thus, of their moving elsewhere." (Tr. 7792-93.) 

The APL line was not to be compatible with the earlier GE 

lines. (R. Bloch, Tr. 7873; see also PX 353, p. 119.) Users of 

earlier GE lines converting to APL would encounter conversion problems 

of the same kind that an IBM user would in converting to APL — and 

GE, of course, would have faced the same problem IBM was confronted 

with in 1964 when it announced 360. It was planned that about 35% 



s 

7 
3 

la i 
u 

13 
15 



L| as primary targets, "the 360/20, 25 and, to some extent, the 30 and, 
Z if to still a lesser, extent, the higher level machines in the IBM line, 
Z\ and also another IBM line". (R. Bloch, Tr. 7663-64, 7866.) GE 
4»| targeted ' these- users because of the difficulties that the users of 
5" | the lower machines in the IBM line would have in converting from DOS 
to OS/360 (R. Bloch, Tr. 7867-68) and the fact that most of the 

programs written for such systems were written in higher level 

languages. (R. Bloch, Tr. 7868-72, 7880; PX 353, p. 64.) 

Bloch testified that he and "the top programming experts" 

at GE believed that the conversion objectives of the APL could be 

achieved, although it was "an extremely ambitious task". (Tr. 

7889-90.) 

To induce IBM's users to move to the new GE systems, a 

price/performance advantage of 20 to 40 percent against the 360 was 

thought to be required, and Bloch thought that it would be necessary 

£g ;| for GE to match IBM's peripherals as well. (R. Bloch, Tr. 7654-59.) 

yj I Basically, what GE was intending to do was to duplicate IBM's 360 

t« :l plan of attack some five or more years after its announcement. 

;i 
t- 'J The strategy for APL preferred by the GE Information 

•? 
„- 1 Systems Group was the "A-F strategy" providing for the offering of 



_? t 



;! shipments beginning in early 1973. This .would have required an $858 



„.. ! million expenditure before taxes, with an after-tax investment of 
m \ \ $429 million for the years 1970-1975. (R. Bloch, Tr. 7695-96; PX 

\ 362, p. 4.) Roughly half of the required investment was the financing 



£• 

?« 



„, : the entire line at once, a "full across-the-board strategy", with 

-1 i • ! 



-531- 



1» 

z 

I 

* 
a 

7 
8 

S 
.0 

.1 ! 

i 

i 

14- !| 
re 

LS , 

L7 [ 
LS 



that would be required for the 80 percent of the APL line that was 
expected to be leased by customers. As Bloch testified, it presumed 
a successful. APL with a large number of systems on lease: "It is 
one of the prices you pay for - success." (Tr. 7699.) "And from my 
view, once those machines were out in the marketplace, we were going 
to keep them out there, which meant tremendous income coming at a 
later time." (Tr. 7929.) " [I]n no sense did I consider 858 to be 
the exposure of the company." (Tr. 7703.) 

Indeed, Ingersoll testified that, during the period when 
he was associated with the Ventures Task Force, it was "a. general 
assumption" that the announcement of APL would have the overall 
effect of increasing revenue and income from GE's current product 
line during the years immediately following 1969, with an increase 
of $177 million from the combined product lines in 1970-1975 (Inger- 
soll, Tr. 8378-82; PX 362, p. 10) , a positive effect not taken into 
account in the $858 million investment estimate. (R. Bloch, Tr. 
7935-36; see also' PX 362, p. 10.) Substantial net profits were 
expected to be earned in the late 1970s, after which a successor to 



•j 

;j the APL was contemplated. (R. Bloch, Tr. 7908-10; PX 362, p. 5.) 

1 The APL plan, then, was an ambitious one, requiring large 

i expenditures. It contemplated an across-the-board attack, even 
21 j 

']• though profits might have been made in a "$50 million business" 

J without such an attack. Further, it had to be pursued immediately. 

y ■» i« 

; In the event, it became just another false start. 

24 f 

I 
t 

1 

•t 

'I -532- 



L 
2. 

<*■ 

4- 

3 

7 
3. 
9 

ia 

Li 

121 

13. 

14. 

15 

16 

I; 

17 l 
19 I 



2a ! 
•i 

'i 
21 ;1 



h. The Ventures Task Force and the Decision to Disengage. 
The Ventures Task Force was formed by GE Chairman Fred J. Borch in 
the last quarter of 1969. It was asked to review GE's computer 
business , commercial jet engine business, and nuclear energy business. 
Its mission was "to analyze those businesses and to present to the 
corporate executive office some plans that would outline the alter- 
nates and options available to the corporation with respect to those 
specific businesses" . (R.. Jones, Tr. 8756-57.) 

In particular, the Task Force stated in its report that 
Mr- Borch "specifically impressed upon us the urgency of our finding 
some way to arrest the heavy continuing drain on our assets resulting 
from these major new ventures". (PX 331-A, p. 5.)* The Task Force 
"adopted two broad criteria as the bases for our efforts to evaluate 
each available strategy; the risks and potential rewards inherent in 
each strategy and impact of each strategy on corporate earnings". 
( Id. , p. 6.) 

Corporate earnings were a problem. GE's earnings per 
share had "plateaued" from 1965 to 1969, creating "a dismal record". 
(PX 331-A, p. 5; see also DX 550, DX 551.) The 1969 earnings had 
declined due to an "extensive strike". (Ingersoll, Tr. 8266-67; see 
also PX 331-A, p. 5; DX 556, p. 3.) As a result, the GE stock price 



j had declined 34% from 1965 through 1969 < compared to a decline of 17% 

for the Dow Jones Industrial Average, 8% for Westinghouse , and an 
1 increase of 3% for the Standard & Poor's Industrials. As the Task 

24 1 

;j Force put it: "Stockholder impatience is indeed understandable." 



.'J * "Major new ventures" were distinguished from GE's "core" 
| businesses. ( Id. ) 

:l -533- 



L 
2 

5 
a 
7 
3 
9 
LO 
11 
•2 
12 



(PX 331-A, pp.. 5-6.) 

The members of the Ventures Task Force, Jones, Jack 
McKitterick, and Robert Estes were corporate officers, but not one 
of them had had any responsibility for the computer business or the 
GE Information Systems Group.* (Ingersoll, Tr. 8267-69.) The 
Ventures Task Force studied the companies that GE met in the market- 
place only "in a superficial way" in order to understand the strategy, 
types of equipment, and the "markets" attempted to be served by "each 
of the major entrants in the business".** (R. Jones, Tr.. 8778-79.) 

Jones testified that the Task Force was a "part time" 
assignment for its members . He characterized the thoroughness or 
completeness of the work done by the Ventures Task Force in the 
following way: 



r . j; "When you look back and think that we worked together for 

" a very limited number of weeks, and when you recognize that the 
computer industry is a very complex business, and when you 
recognize also the fact that none of us had any experience in 
, „ 'j the computer business before we went into this, certainly it 
* B • was not an exhaustive analysis of the computer business. It 
t-> ; was an analysis that I think developed a fair comprehension of 
^ y \ General Electric 's position in the computer business, but I 

wouldn't characterize it as an in-depth study." (Tr. 8767-6 8.) 






Of the three new "ventures" — computers, jet engines, and 



\ nuclear energy — the Ventures Task Force studied computers first. 

1Q J 



21 i 



I * At the time of their Task Force assignment, Jones was Vice 

22 ] President of Finance, McKitterick was Vice President of Corporate 

:| Planning and Estes was Vice President, Secretary and General Counsel 

23 \ (Ingersoll, Tr. 8267-69.) 

2- t ** The Task Force studied IBM, NCR, CDC, Honeywell, Burroughs, 

•I Univac, Xerox, ICL, Siemens, and several JaDanese companies. (R. 
25 I Jones, Tr.. 8778.) 



-534- 



I 

2 : 
3 

a 
6 

i 
3! 

i 

10 | 

11 ! 

12 I 

13 ! 

14 I 

i 

15 ' 

16 i! 

17 \ 

IS ; 

19 : 



This was because, as Jones testified, GE was "in a position in 
nuclear where we had so many contractual commitments that our options 
and our alternates were rather restricted. The same thing held to a 
lesser degree in the commercial aircraft engine business, whereas in 
the computer business it seemed to us that we had a good deal more 
freedom to select from a rather wide range of alternates and options 
as to the future course of the business". This was so because the 
computer business "did not have long-range, long-standing contractual 
commitments to deliver product [s] over an extended period of years". 
(Tr. 8758-59.) 

The Ventures Task Force "attempted to evaluate the risks 
associated with the APL plans . . . from a broad business standpoint. 
. . . [It] did not undertake to verify the accuracy of specific 
details of the cost estimates, for example." (Ingersoll, Tr. 8431.) 

The Task Force ultimately reached the conclusion that GE 
should "disengage by combining its computer business with that of 
some other computer manufacturer" (R. Jones, Tr. 8801) despite the 
fact that it found that in the computer market "great size and very 
rapid growth make for a challenging opportunity" with the U.S. and 



„ ,1 European businesses projected to double in the next five years. (PX 

£U i 
:l 

;i 331-A, p. 9.) It listed a number of negative factors affecting GE:. 

•I "substantial operating losses", "heavy debt obligations and interest 
22 



23 
24 



burden", "obsolete product lines", and "poor reputation and image" 
j (PX 331-A, p. 49), and stated that GE had 



;| "Limited technical strength other than in data management and 
25 ;j multi-processing software and communication equipment. 



'! 

";! -535- 



I I "Major product lines obsolete, complete but incompatible. 

Not vertically integrated. Weak in peripherals, mass storage 
Z\ and terminals." (PX 371-A, p. 39.) 

3 | Among listed "Critical Future Problems" were 

i . 
4 | "Across the board system obsolescence. 

5 J' "Vulnerability of PARC [installed base] to competition — lack of 

specialization. Customer loyalty now under 80%, lowest of any 
$ ; competitor. 

i 

7J "Lagging technical position in main frames, peripherals, and 
manufacturing process technology." ( Id. , p. 40.) 

3 ' 

9 
10. 
11 
12! 
13 
14 
15 

16 :! 

■\ 

17 ' 
u •■'. that plan 

* t " conceptually recognizes the current needs of the business and 

Tq 1 presents a goal that, if realized, would indeed place the 

:| company in a strong position in the business computer field. 

:j It is our conclusion, however, that the APL entails very high 

*° :| risks, and that it is doubtful that it could be kept to time, 

21 ;j 

ij 

22 :j * Jones also testified: " [I]t is my experience that in business 
«- :! you succeed when you satisfy a customer and when you do it in terms 
2-* ;j of giving values that are highly satisfactory from the standpoint of 

the customer. And I use 'value' in the sense of conveying reasonable 
price, quality of product, features of product and performance, 
overall performance of product." (Tr. 8 868.) 



Its installed base was termed "already obsolete and vulnerable with 
the conclusion: "Time is not on our side." (PX 331-A, p. 20.) The 
Task Force stated that "we need to be realistic about the relatively 
poor reputation and image we enjoy as a computer equipment manufac- 
turer". (Idw p. 34.) 

Jones concluded: " [W] e were not doing the job that was 
satisfying the customer to the extent that certain competitors 
were." (Tr. 8886-87.)* 

The Task Force evaluated the APL plan. It concluded that 



24 
25 



-536- 



I 

a! 

I 

i 

4 | 

si 

i 
I 

5 \ 

7| 

8'j 

i 

*! 

J 

lfl i 

11 j 

12 j| 

i 
i 

13 I 
14 
15 i 
16 



cost and system performance schedules. Even if General Electric 
were in a position to undertake such an ambitious program, we 
would not recommend that it invest the requested sums in such a 
hazardous project predicated on an all-out attack on IBM, one 
of the world's strongest corporations. 

"Faced with the lack of earnings growth, but seeking to 
retain its image as a growth company, General Electric cannot, 
in our opinion, undertake any half-billion dollar venture, such 
as APL that produces substantial immediate net income losses."" 
(PX 331-A, p. 7, emphasis in original.) 

The APL plan, according to the Task Force, called for a fourfold 
expansion in GE total shipments in six years with an expansion of 
60% to 70% per year of its sales force and "even so, productivity of 
GEC's] sales force must be twice as great per man as that of IBM". 
( Id. , p-. 28.) 

Jones testified that one of the reasons that the Ventures 
Task Force "felt that the APL plan was one fraught with risk" was 
that it called for technology beyond the current state of the art, 
which required "invention by schedule in order to achieve its objec- 
tives". (Tr. 8769.) If it "had not been so ambitious in a techno- 



i 
17 j| logical sense and in a timing sense it might have been a somewhat 

l' it 



IS 
19 
20 
21 



I better plan. It might have had a chance of acceptance." (R. Jones, 

i 

I Tr. 8790.) 

i 

| In fact, Weil testified that by late 1969, APL was not 

i 
I 

| "well along" in its design and development and that "the software 



!{ was still in fairly early specification form". (Tr. 7244-45.) At 
ZZ : : 



23 
24 
25 



the time of the Ventures Task Force "detailed engineering specifica- 
tions" and "firm cost estimates" were not available. (Ingersoll, 
Tr. 8370; see also R. Jones, Tr. 8768; PX 363, p. 15.) 

Yet the world would not wait for GE to come up with its 



-537- 



1 
1 

3 

5 
5 
7 
8 

10 
11 
12 
13 
14 
15 
16 
17 
IS 
19 
20 



360. Jones testified that "it was our experience that every time we 
went out to sell a computer there were a lot of other people knocking 
on the customer's door, attempting to sell him a computer. In that 
sense it was highly competitive." (Tr. 8861.) He testified that 
"it was the opinion of the Ventures Task Force that some of the 
companies in the field would find it necessary to combine . . . that 
the [business computer systems] business was competitive . . . and 
that it would continue to be competitive and it would be those 
competitive pressures that would cause some combinations to take 
place in the field." (Tr. 8864-65.) General Electric believed 
"that one of the characteristics of the business computer systems 
business was that competition constantly forced suppliers to come 
out with better products at lower prices in order to keep the custo- 
mers that they had and to get new customers". (R. Jones, Tr. 8865.) 
The Ventures Task Force called IBM "a moving target" which 
Ingersoll interpreted to mean that "we at General Electric should 
assume that IBM would not be a stationary object, it would be a 
dynamic situation, and the conditions . . . might well change". "[I]t 
was in effect a high risk to assume that the conditions and evalua- 
tions . . . would remain constant, that is, that the comparisons 



•i 

•{ would be subject to change as IBM made plans and introduced its own 

21 (j 

|l products." (Ingersoll, Tr. 8128-29; PX 363, p. 15.) Similarly, the 

22 :j 

i Ventures Task Force and its support staff felt that "it was a high 

23 | 

■ 

I risk assumption" to assume that competitors other than IBM "would 
24 

; stand still with respect to market share". (Ingersoll, Tr. 8127.) 

: \ 

'i 

i -538- 



L 
Z 
3 

5 

7 

i 

3 ; 

-I 

ii i 

13 I 

! 

i 
15 ! 

is ; 



As a result, Jones testified that the Ventures Task Force 
concluded "that the life of a family of computers was quite limited", 
something in the range of four to six years, "and that you did not 
bring out a family of products that simply met the price/performance 
characteristics of the then existing competition. You had to bring 
out something that would exceed the price/performance of the exist- 
ing competition because you knew full well that they were going to 
be moving ahead of you. It is a constant leap frogging game." (Tr. 
8866-67.) It was a lesson which GE learned too late and IBM had 
learned well before. (See PX 1077;* DX 4806;** pp. 493-94, 531 above 

Indeed, the APL plan itself stated in January 1970: 

"One of the key aspects of technology in the computer field is 
its high rate of obsolescence. Never in the history of tech- 
nology has the pressure of competition and the lure of highly 
rewarding markets created such a dynamic evolution. While this 
characteristic is forcing rapid technological progress, it is, 
at the same time, imposing on the computer manufacturer a heavy 
financial burden and the necessity of planning products with a 
narrow margin for error." (PX 353, p. 23.) 



The Ventures Task Force, of course, had not been called 

•' jj together simply to consider computers. Ingersoll testified that he 
IS 

19 

20 



!( 



* Thomas J. Watson, Jr.: "I believe that whenever we make a new 
j machine announcement, we should set up a future date at which point 
| we can reasonably assume that a competitor's article of greater 
'•« capability will be announced. We should then target our own develop- 
2 * 1 ment program to produce a better machine on or before that date." 
:j (PX 1077, p. 2.) 



-« ;, ** Thomas J. Watson, Jr.: "I think it important to note, however, 
2,3 !J since we seem to have suffered for a few months or even years because 

;i our machines predated the effective competitive machines now in the 
2** ;| marketplace, that we now make these machines good enough so they 
j will not be just equal to competition, for I am sure that once they 
j are announced our competitors will immediately try to better them. 
j This is all to the good and I am for competition, but I want our new 
j line to last long enough so we do not go in the red." (DX 480 6.) 

-539- 



25 



I ! 



believed that GE's management was concerned with improving the 
profitability of the company as a whole, based in part on the expe- 
rience of the late 1960s in terms of stationary and then declining 
earnings. It was therefore concerned with the immediate impact of 
APL on GE's earnings which impact, from 1970 to 1975, the APL plan had 
projected to be negative. (Ingersoll, Tr. 8271-72.) 
r !( The Ventures Task Force concluded that "General Electric 

3 \ can ill afford the financial resources needed for an all-out drive 
a for position in this industry, basically because of the needs of 
. i other businesses within its scope". (PX 371-A, p. 73.) The "core" 
businesses needed "more rather than less support, and the company's 
immediate earnings goals can only be met from these businesses". 



1 I 
1! 



(Id., p. 76.) 



.1 



As Jones testified: 

"At that point in time we had these two tigers by the tail 
other than computers — that is, jet engines and nuclear — and we 
^ |! had this host of other ventures that we were trying very hard j 
■Q ! to bring on, all of them spun out of this common technology of 

*i our electrical industry and electronics industry. 
.7 i| I 

"We were increasing or had been increasing our debt subs tan- j 
.3 ;j tially through this period, so that our debt-to-capital ratio 

| has been climbing. And we just said, you know, there is a 
.3 | breaking point where we will lose our triple A rating as a j 

I corporation if we continue to pile on debt and if we continue 
10 j to try to do all these things that we have got on our plate j 
right now. I 

j "This was the one where we had you might say the least j 

H > commitment ... in terms of contractual commitments." (Tr. 

8831-32.) j 



12 



• The Ventures Task Force also had concluded that: 

2- I 

j "For the first time in our generation, at least, we face the 

?5 .j necessity for an allocation of corporate resources which are 



; ! -540- 

i 
i 

! 



I 

2 

Ji 
i 

«•! 

si 

1 

i 
i 

T\ 

i 

I 

s I 

i 

»! 

12 j 

13 | 

i 

i 
i 

15 i 

i 

16 I 

17 i 

:t 

1S;| 

19 i 

20 :{ 

21 | 
22 
23 
24 
25 



not adequate to meet all of our readily identifiable needs — 
during a period when the company is under special pressure to 
demonstrate its ability to grow earnings. The general economic 
climate is not favorable; the capital markets are severely 
depressed; credit is costly and may be assumed to become pro- 
gressively less available;- inflation has forced higher labor 
costs on the company following the longest strike in the company's 
history." (PX 371-A, pp. 75-76, see also pp. 7, 10; PX 331-A, 
P- 7.) 

GE could not do everything at once: "[W]e can't afford to 
back every horse. We just don't have limitless resources." (R. 
Jones, Tr. 8843.) 

Disengagement in nuclear energy and jet engines was "not 
an available option" because of GE's contractual commitments. (PX 
371-A, p. 76.) While nuclear reactors were considered part of GE's 
"core business" ("essentially those that dealt with the generation, 
transmission, and distribution of the electrical energy in terms of 
the equipment to do all those jobs, plus the equipment that would 
utilize electrical energy") and jet engines were considered "a spin- 
off of the core, but . . . very closely related", General Electric 
never "viewed the computer business as being part of its basic 
core". (R. Jones, Tr. 8838-41.)* 

The Ventures Task Force did consider the possibility of 
retrenchment, rather than withdrawal, with such retrenchment taking 
the form of moving to a more specialized product line. However, it 
was decided that this was not an optimal strategy for General Electric, 



* It is instructive that in its "core business", nuclear reactors, 
GE had invested for 20 years without making a orofit. (Ingersoll, 
Tr. 8288-91.) 



-541- 



t 

r 

\ 
} 
J 

L 
Z 

z 

£ t 

7\ 
,S ! 
.9 
>0 



although it would have required less investment money. (Ingersoll, 

Tr. 8144-45, 8150-51, 8188-89; R. Jones, Tr. 8801, 8881-83, 8796- 

97.) As Jones testified, this was in part because: 

n Vte had sold our equipment to almost every market. We 
had not concentrated on banking or manufacturing or retailing. 
We had either of our own manufacture or through resale a sub- 
stantial and wide ranging line of peripherals, and because our 
product offerings over time had been so eclectic, we felt that 
if we were to withdraw to the role of a specialist, we would be 
in effect abdicating many of the customers and many of the 
markets that we had been serving. 

"Of course, the result of that would be a substantial 
reduction in our overall market opportunities." (Tr. 8797.) 

The decision was made to merge a large part of the business 
with Honeywell by both companies transferring their business computer 
operations to a new Honeywell subsidiary. (DX 555, p. 3.) GE retained 
an interest in the new company and, as is detailed below, success- 
fully continued its "own independent development of businesses in 
the promising areas of process computers, computer time-sharing and 
data communications equipment."* ( Id. ) (See pp. 544-46 below.) 

As we have seen, the GE decision to disengage was taken in 
part because of past GE mistakes and in part because General Electric 
!( had other fish to fry.** 



;! * GE would have liked to have taken a controlling interest in the 
l| new venture but feared the disapproval of the Antitrust Division. 
1 (See Ingersoll, Tr. 8242, 8252-5 3; DX 7259, Borch, pp. 13-14.) 



J ** Jones testified that he "personally knew of no acts or activities j 

22 :| of IBM that would have caused our disengagement", and none were j 

| brought to his attention during the activities of the Ventures Task j 

2^) Force. (Tr. 8867-68.) ! 

:! i 

25 : ; 

i 

i 

j -542- I 

i ! 

•»l . ■ j 

; l ! 



L 
2 

3 

f 

a > 
7 | 
3 !| 
9- 

10 

11 

12. « 

i 

13 ; 
14 
15' 
IS 



17 ! | 



13 



Could General Electric have succeeded had it remained? 
Bloch testified that he thought GE had the resources necessary to 
become "a clear No. 2 in the supply of computer systems used for 
business data processing purposes". He ,r felt at the time that it 
■was a mistake" not to implement the APL. (Tr. 7811.) Withington, 
who, unlike Bloch, was not personally involved, testified that GE 
had sufficient assets to be successful if it chose to invest them in 
the general purpose computer systems business. (Tr. 56732.) 

Indeed, General Electric did not have much trouble raising 
money in the years after 1970. From 1970 to 1974, it raised approxi- 
mately $625 million by long-term debt. (DX 553.) It chose to 
invest those funds elsewhere. 

i. Did GE Lose Money? It is questionable whether GE lost 
much, if any, money in the course of its computer operations or, at 
least, whether it would have lost money thenceforward had it not 
sold part of its operations to Honeywell. 
I A report- by Peat, Marwick & Mitchell, General Electric 's 

utside auditors, showed that GE ' s domestic business computer opera- 
tions lost approximately $163 million in the period 1957 through 



19 i 

■{September 1970. (PX 380; see Ingersoll, Tr. 8353-56.) Jones testified 

20.1 

jthat this accorded with his recollection. (Tr. 3756.) Ingersoll 

21 =1 . 

testified that this figure included allocation of corporate overhead 

22 '"> 

expense to GE ' s domestic business computer operation, involving 

23 | 

expenses which continued after the transfer to Honeywell and which 

v-\ 

yere not directly incurred in the operation of GE ' s domestic computer 

\\ 



-543- 



i 

3 i 



operations. (Such allocation amounted to 10% to 15%.) (Ingersoll, Tr. 
8359-60 , 8365-66.) 

Further, Ingersoll testified that PX 380 "reflects the 
cost of developing the equipment incurred by the Domestic Business 
Computer Operations that were subsequently transferred to Honeywell— 
the development/ that is, through the date of the transfer". (Tr. 



j\\ 8377-78.) 



3 

3 

a 
i 

2 
.5 



Hence, in order to evaluate whether GE would have ultimately 
made losses had it not sold part of its operations to Honeywell, one 
must consider the profit stream which would have resulted from 
those sunk expenses. As Ingersoll testified, "because of the rela- 
tively long life of computer equipment, ... in order to properly 
evaluate the total results one should look toward the full life of 
the equipment" . He agreed that it is "generally true that in the 
development, manufacture and marketing of new lines of computer 
• equipment that losses are sustained in the early years and profits 



;| derived in later years". (Tr. 8199-200.) 

" !| 
!| According to GE's proxy statement, that portion of GE's 

• s J 

:| computer operations which was sold to Honeywell had an after-tax 

j profit for 1969. (DX 554, p. 9; see Ingersoll, Tr. 8329-30.). 

10 \ 

I Further, in December 1969, GE's Information Systems Group estimated • 

11 : ! 

j that the net income from its current computer product lines for the 
22 '} 

.] years 1970 to 1975 would be $173 million (this estimate also included 

2 I 

a positive impact of APL on current products). (PX 362, p. 10; see 

24. 1 

j Ingersoll, Tr. 8378-79.). The Ventures Task Force in April 1970 

5S I 



• -544- 

i 
i 

.1 
.j 



I 

2 ! 

3J 

3 



a 



estimated that the current product line would bring in $821 million 
in revenue and $164 million in net income in the years following 
1969, regardless of APL (it was evaluating the business from the 
point of view of a prospective buyer) . (PX 331-A, p. 32; see 
Ingersoll, Tr. 8376-77.) 

~ lj The terms of the sale, which was announced on May 20, 1970 

jt (PX 323 (DX 14502)), were as follows. The two companies formed Honeywell 

fl r Information Systems (HIS) and GE received an 18-1/2 percent interest 

3 ! 

in it. In addition, GE received 1.5 million shares of Honeywell 
common stock and $110 million of Honeywell subordinated notes (later 
converted to additional shares of Honeywell common). (DX 555, pp. 
22-24; see Ingersoll, Tr. 3393-96; DX 14073, p. 32.) At the time GE 
recorded a profit of $1.7 million on the transaction. That amount 
was quite conservative, GE having undervalued its minority interest 
and the value of. its Honeywell stock. (DX 555, pp. 26, 31.) Ingersol] 
testified that the market value of the 1.5 million shares of Honeywell 
stock received by GE was "in the neighborhood of $130 million". 
(Tr. 8388-89.) In addition to that and the $110 million in notes, 
Ingersoll testified that GE valued its minority interest in HIS at 
"approximately $32 million", "substantially less than the net book 
value of that minority interest as determined by Honeywell" which 
valued it at "at least a hundred million dollars". (Tr. 8393-96.) 
In fact, the 1,500,000 shares of Honeywell stock had an average 
market value of about $120 million as of the last quarter of 1970, 
based upon the average between the high and low prices at which it 



II 
12. 



13 



14 
15 
IS 
17 
IS 
19 

2a 

21 
22 
22 
24 
25 



-545- 



^ traded during that period. (DX 555, p. 31; DX 14064, p. 164.) In 

2 1971, GE received 1,025,432 shares of Honeywell stock in exchange 

3 I for the $110 million in notes. (DX 14073, p. 32.) Those shares had 
a ; an average market value of about $113.2 million in 1971. (DX 14-130, 
5 • p. 164? DX 14131, p. 172? DX 14132 r p. 172; DX 14133, p. 173.)* In 

g ; 1976, GE exchanged about one-third of its interest in HIS for 800,000 

j shares of Honeywell stock and, in 1977, the remaining two-thirds of 

g another 1,400,000 shares. (DX 13980, p. 40; DX 13981, p. 40.) The 

g ' Honeywell stock received in 1976 had an average market value in that 

1Q ( year of about $35.6 million (DX 14062, pp. 2, 4, 6, 8); the stock 



11 



received in 1977 had an average market value in that year of about 

12 $68.5 million. (DX 14063 r pp. 2, 4, 6, 8.) GE sold the Honeywell 

13 stock over the years, disposing of the last in 1978. (DX 13981, p. 

14 40; DX 13887, p. 40.) 

15 Taking all these things into account, even allowing for 

lg the difference in timing and inflation between the expenditures made 

27 . in the early sixties and the returns received from the Honeywell 

lg , sale, GE appears to have been a net gainer in the computer industry. 

19 
20 
21 
22 
23 



24 * As used herein, the average price of Honeywell stock in a 
given year is the average of the high and low prices at which it 

25 traded during that year. 

-546- 



L || 42. RCA. As we have seen, the 1950s were a stagnant 

i ' 
Z \ period for RCA' s computer business. During the early 1960s RCA 

^ : | experienced several problems that continued to retard its growth in 

4* [\ computers, in particular the failure of the 601 and its on again-off 

again peripherals development. (See pp, 196-207 above.) Toward the 

5 f middle of 1963, RCA had stopped marketing the failed 601; the 501 

jl 
7 i| was "starting to decline" and RCA was marketing only one computer 



3 

io : 

13 
14. 



model, the 301. (McCollister, Tr.. 9622.)* RCA then announced its 
3301 computer system. 

a# RCA 3301. The RCA 3301 was announced on August 20, 
1963. (DX 580, p. 1.) The 3301 was an "interim product", designed 
and marketed by RCA "to round out our overall product program . . . 
[by] tak[ing] the place of the 601". (McCollister, Tr. 9247, 9622.) 
The 3301 



"was not a new design. It wasn't intended to be the 
foundation of a future line of products; rather, it 
T c ': was a product that we could develop relatively quickly, 

* ;; at relatively low engineering expense, [**] that would 

17 | 

18 i 

TCI 

• -j * E. S. McCollister joined RCA in 1961 as Vice President of 

II Marketing for the computer division. He held that or a similar 
- a .| position until December 1971 when he left RCA. He then joined the 
j Burroughs Corporation in January 1972. (McCollister, Tr. 9161-62.) 

j ** McCollister recalled that the engineering development cost 
22- -j for the 3301 was "in the order of about $'2 million for the 

:j processor and for the associated control units". (McCollister, 
22 Tr. 9623.) 

2*1 

i 

25 1 



-547- 



give us an additional offering to take the place of the 
601, and that in a sense would give us time to get on 
with a complete new product program in the longer 
range future." (MeCollister, Tr. 9622-23.) 

RCA described the 3301 as an ,r all purpose computer" that 

"features advanced communications devices and arithmetic circuitry 

to make it equally powerful for scientific equation solving, super 

fast business data processing, instantaneous (real-time) management 

control, and high-speed data communications". (DX 580, p. 1; see 

also Beard, Tr- 8994-95.)* In addition, the 3301 could perform 

applications "which were in other circumstances performed by multiple 

installations of special purpose computers". (Beard, Tr. 8994- 

95.)** 

While a more successful product than the 601, the 3301' s 

success was limited for two reasons : poor peripherals and its 

"eclipse" by the announcement of the Spectra 70 series less than a 

year and a half later. (Beard, Tr. 8458-60, 10276, 10307.) 

1 : RCA had been beset by problems with the peripherals used 



M 



r 



X 



on its 501 and 301 systems. (See above, pp. 195, 201.) The peripherals 
on the 3301 "prevented the computer system from achieving its full 



* : j throughput capabilities". (Beard, Tr. 10276.) Although RCA had 
j 

^ ;j resumed manufacturing its own peripherals in 1962 (see above, p. 



T I 



~ j * The 3301 was not as versatile, however, as System/360. (Beard, 

- ! Tr. 10266.) 

I 
A j ** A. D. Beard, Chief Engineer of the RCA computer division from 
~ j 1962 until 1970 (Beard, Tr. 3447-51) , used the term "special purpose 

- | computer" to mean "real time" computers such as SAGE (Tr. 8995-97), 

~ | "communication equipments", and "small and medium sized scientific 

j computers". (Beard, Tr. 8996.) 

i 

1 

■I 

': -548- 



I 

2 



3 j, existing on the 301 or which could be made available from outside 



4- 
f 

7 I 

3 

9- 

10 
11 
12 
13 
14- 
15 



202) , time and continuity of effort had been lost and RCA was 
largely "constrained to live with the peripherals that were then 



suppliers". (Beard, Tr. 9004, 1.0307; see also McCollister, Tr. 

9622-23.)* 

RCA also experienced problems with a peripheral unit of its 

own manufacture, the RACE mass storage unit. The RACE unit was a 

storage device that used magnetic cards . ** Those cards 

"had to be extracted from a magazine, put in a channel that 
carried it to a revolving drum, held on the drum while it 
rotated past a reading head, where the information was read or 
reported, and then the card had to be returned to the magazine 
from whence it came. 

"And this was a very, very complex mechanism and a very 
difficult technical task." (McCollister, Tr. 9657.) 



RACE was designed to provide random access storage for the 
3301. Compared to IBM's 2311 disk drive,/ announced by IBM with 
System/360 in April 1964, RACE was "much smaller" in terms of storage, 
15 :! but "considerably faster" in terms of access time. Thus, for the 
*7 !| application mix of some users, RACE, when operating properly, would 
-S i| be superior to a disk drive and under other circumstances , the disk 

19 j 

11 

20 '| * According to McCollister, "In the 3301 [RCA] used a card reader 
I from Uptime, we used the ICL, or ICT I believe it was at the time, a 

2- ;j hundred-card-a-minute punch, we used the Anelex printer, of which we 

i bought the complete printer sub-system from Anelex." (Tr. 9622-23.) 

22 '■'> 
•[ 

ij ** The RACE unit came in two models, the 3488 to be used with the 

23 | 3301 processor and the 568 for the later Spectra series. (McCollister, 
ITr. 9656-57.) 

24 'I 

;! / RCA did not have a disk product of its own to offer with the 

25 ;j 33-01. (Beard, Tr. 9046.) 



-549- 



drive was superior. (Beard, Tr. 9046-47.) 

The major problem with RACE was that it was not reliable. 

Thus, Wlthington classified RACE as "a major product failure" (Tr. 

56511) , and McCol lister explained why: 

"[Tlhe cards wore out . » . the cards were damaged in transit 
. . . sometimes there was a failure to select the proper card 
... it was a tedious process to replace a card in the file 
when it was beginning to wear out and, indeed, to detect when 
it was beginning to wear out." (Tr. 9658.) 

Moreover, even when operational the RACE unit was "unable to meet the 
speed of accessibility that had originally been specified in the pro- 
duct" . (McCollister, Tr. 9658.) 

Second, the success of the 3301 was limited by RCA's intro- 
duction of a new series only a little more than a year after the 3301 
was announced. RCA's announcement of its Spectra 70 series in 
December 1964 "eclipsed" and "superseded" the 3301. (Beard, Tr. 
8458-59; PX 4830, p. 25.) Thus, potential customers of the 3 301 were 
encouraged to obtain Spectra 70 series. For example, RCA provided 
emulation of the 301 on the Spectra 70 but not on the 3301. (PX 
4830, p. 25.) Another sign to RCA's users that the 3301 was a dead- 
end machine was RCA's failure to provide a "growth machine" for users 



3 | 

\ of the 3301.* (Beard, Tr. 9986-87.) Because RCA provided no emulation ! 

3 i i 

i from the 3301 to Spectra 70 or to any other system, 3301 users had ; 

3 ' I 

2. ! l i 



3 s * a "growth machine" is one "which allows the programs to be moved 
i from the predecessor machine to the growth machine [with] a minimum 
.4 ! of re-programming effort". (Beard, Tr. 9986-87.) 



■« 



-550- 



nowhere to go in the RCA line without converting their programs. 

2 j (Beard, Tr. 8458-59, 10235.) 

b. The Spectra 70 Series. IBM announced its System/360 

4-j on April 7 , 1964. Beginning "shortly" after the announcement, RCA 

formulated the "design specifications" for its Spectra series. Those 

5 ';! specifications were done "in [a] preliminary fashion" around July or 
August of 1964. (McCollister, Tr. 9624.) 

The Spectra 70 series eventually comprised eight models — 
the 70/15, 70/25, 70/35, 70/45, 70/46, 70/55, 70/60 and 70/61. The 
sizes of the processors increased in numerical order, and the 70/46 
and 70/61 were intended to offer time-sharing capabilities. In 
December 1964 RCA announced the 70/15, 70/25, 70/45 and 70/55. 
(Beard, Tr. 8483-85; McCollister, Tr. 9635-36; DX 669, p. 11.) The 
70/35 was announced in September 1965 (DX 670, p. 16); the 70/46 was 
announced in 1967 (PX 338, p. 22); the 70/60 and 70/61 were not 

announced until 1969 (DX 674, pp. 8-9) . 

la |i 

j No prototype of any of the systems was in existence at the 

I time of the announcement. A prototype of the first machine was not 
IS :| 

j built until the middle of 1965 at which time prototypes of most of 

19 i 

ir the control units were also built. (McCollister, Tr. 9635-36.)* 



7 s 

a ; 

9 j 

ia j 
ii j 

12 j 

^ \ 
L4 ; 



?T 



! * During 1965 through 1966 several new peripherals were also 
— :• designed. (McCollister, Tr. 9635-36.) During 1965 and 1966 RCA's 



i total engineering budget for software and hardware was approximately 
-^ j 315 million annually, out of which came engineering expenditures 
mjk -j for the support of older products as well. (McCollister, Tr. 9634-35.) 
-" j Compare the status of RCA's Spectra at announcement with that of 

I IBM's System/360: 3y the time of the IBM 360 announcement there 
- 3 j were prototypes of all models of the processors. (3rooks, Tr. 

•j 

■! -551- 



Deliveries of the "small systems" began in 1965 and the "larger 
systems" in 1966. (Beard, Tr. 8460.) 

Four aspects of the Spectra 70 series are particularly 
important: its attempt at compatibility with System/360, its ability 
to perform commercial and scientific applications, the problems RCA 
encountered and the success of the Spectra 70 series. 

(i) Compatibility with System/360 . In a decision that 
affected both the Spectra 70 and its succeeding RCA Series, RCA 
decided to make its Spectra 70 series compatible, that is, able to 
use the same application programs -with little or no modification, 
with IBM's 360 systems. 

By making its Spectra 70 compatible with IBM's System/360 
RCA hoped to be able to persuade substantial numbers of 360 users to 
move to the Spectra series. (Beard, Tr. 8461-63; McCollister, Tr. 
! | 9269-70; Rooney,* Tr. 12117.) In particular, RCA expected to target 



<n 



7 ! 22695-6; Hughes, Tr. 33995.) Most of the processors and some of the 

! j peripheral equipment were in the early stages of product test 

3 .J (McCarter, Tr. 88382-83; JX 38, 1! 19); all, or almost all, the 

'I memories had undergone technical evaluation testing (Brooks, Tr. 

3 \ 22699) ; microprogramming and multiprogramming had been tested on 

; j the Model 40 (McCarter, Tr. 88382-83) ; four estimating, forecasting 

£ I and pricing cycles had been completed; and the "componentry, 

I systems and product testing program already completed . . . [was] more 

LI '■ extensive than the entire program ever [previously] undertaken for a 

3 system". (DX 1172, p. 2; see also DX 1165.) 

j * J. W. Rooney joined RCA as Vice President for Marketing Operations 

23 in 1969. He became the Vice President of Marketing for the Computer 

I Systems Division in 1970 and was President of the Division from 1971 

Ml until he left to go to Itel in 1972. (Rooney, Tr. 11687-88.) 



j -552- 



L 
2 
3 

a 
7 
3 
9 

LO 

11 
12 - ; 
13 
14- 
15 
IS ; 



;i 



n \ 



its marketing efforts at those 360 users who wanted to obtain 
larger or more functional equipment. (Beard, Tr. 85 26-28.*) To 
those users RCA wanted to offer better price/performance on its 
Spectra 70 equipment than IBM did on its 360 equipment and thus 
persuade the user to acquire Spectra 70 equipment. (Beard, Tr. 
10103; Rooney, Tr. 12117.) 

Thus, during this period RCA attempted to offer a price/ 
performance advantage of between 15% and 20% over IBM's systems. 
(Beard, Tr. 8493-94, 10095; Rooney, Tr. 11826; see also Wright,** 
Tr. 13083.) RCA's pricing methodology was the direct result of 
RCA's strategy to be compatible with IBM and to go after IBM users. 



* RCA believed that compatibility had become a more useful 
marketing tool after introduction of 360. Prior to 360 RCA had a 
"sales opportunity" whenever a customer wanted to go to another 
computer because: 



"As the computer industry evolved, ordinarily even 
in moving within the line of one manufacturer, within 
;| the line of IBM, there was conversion that was necessary. 
17 ;{ And since a customer was facing ... a conversion in the 

: case of IBM, we could argue it won't be any more difficult 
13, to convert to RCA." (McCollister, Tr. 9273-77.) 



IS ; i Since users of 360 would be able to move up to a larger IBM 
"I 360 computer "relatively easily", it would be to RCA's 

20 j advantage "to make it almost as easy as we could for the cus- 
tomer to move to the RCA product line as to move up within the 
IBM product line". (McCollister, Tr. 9273-75.) 

** V. 0. Wright joined RCA as Vice President and Regional Manager 

of Federal Government Marketing in the Computer Division in 1970, 

became head of Systems Development in 1971 and left in 1972 to go to 

! Amdahl. (Wright, Tr. 12785.)" 

24 ! 
i 



-553- 



"[I]f you are going head to head with a competitor, 
such as IBM, and you essentially are going to offer the 
same function and if the competitor is in a stronger 
position in the marketplace than you are, you would 
offer a price advantage to move your product." (Rooney r 
Tr. 12415, see also Tr. 12414, 12420-21.) 

The pricing methodology was based upon two assumptions: 

(1) RCA assumed that in many cases Spectra would be 
offered to displace existing IBM computers and some 
inducement would have to be offered to persuade the 
IBM user to go to the trouble of replacing his existing 
IBM computers and install Spectras (Beard, Tr. 10103; 
Wright, 13083-84)*; and 

(2) Spectra was delivered one to two years after 
IBM's System/360, and customers expected a new offering 
to have a price/performance advantage over older computers . 
(Beard, Tr. 10103-05.) 

If RCA had not adopted the compatibility strategy, other pricing 

strategies would have been available: 

' i "If you did not have compatibility, you would be 

going with your own product line, which would have its 
3 ; t own unique characteristics and functional capabilities. 

s\ 

a | 

:i * This "was needed in order to compete with IBM and in 

— j order to obtain business from people who were currently using IBM 

'j systems, to displace IBM systems. If a customer has a fairly 

£ ;| substantial investment in software, he has to have some reason and 

j some motivation to move to another vendor, and we felt that that was 

S | the margin of motivation that was needed to get them to move." 

| (Wright, Tr. 13083-84.) 



i 



15 



e ■ 

:i 



-554- 



L 

2 ! 

i 

i 

5 : | 

I 

& i 

7 
3 

10 

II 
12 
13 
14- 
15 
la 
17 
13 

•j ". . . [Tjhere would be some possible configurations 

2Q ] whereby you would not meet your price/performance goal 

:| of 15 to 20 percent; you would only meet them on the 

2£ j specific points that you had evaluated. 

22 •; "There would be some cases where you would 

j exceed that price/performance advantage. There would 
~- ;j be other cases where perhaps the advantage would go 

24 1 

I 
•( 

] -555- 



"If you felt that you were matched in those unique 
capabilities and functions by IBM, you would probably 
price your products for comparable systems in the same 
range but not necessarily under IBM. 

"If you felt you. had a unique product, then you 
would not be guided by IBM's pricing policies. You 
would then price based upon your analysis of what the 
market would bear for that functional capability." 
(Rooney, Tr. 12421.) 

RCA employed the technique of "straddling", placing its 

machines in terms of performance "approximately midway between a 

pair of IBM machines". (Beard, Tr. 10106-07, 10113, 10121-22, see 

also Tr. 10097-99.) 

"Generally what was done was to pick what 
appeared to be the most commonly used configurations 
of equipments : so many tape drives for a small 
system, so many tape drives for a large system, and 
so forth, to pick several of what were considered to 
be representative points around which you would expect 
a large number of customers to cluster. 

"Based on that simplified set of system con- 
siderations — that is, not taking all the possible 
combinations into account, but some of the most probable 
ones— evaluations were made on a system basis, that is, 
not only the. speed capabilities of the main processor 
but also what range of peripheral speeds you would put 
on each of these system configurations , determine what 
the relative overall performance advantage or dis- 
advantage was, and set the prices accordingly. 



the other way." (Beard, Tr. 10092-93; see also Rooney, 
Tr. 12129-31.)* 

As discussed at greater length below, RCA's attempt to offer 
better value to the customer than System/360 failed even though it 
may have been able to announce that it provided more throughput per 
dollar (better price/performance) than some of the 360 systems with 
which it competed. 

RCA also believed that the compatibility strategy would 
have a good chance for success because it could take advantage of the 
situation where IBM users were leasing from IBM on a short-term 
basis, which leases could readily be terminated and RCA equipment 
substituted. (Beard, Tr. 10073; see also Rooney, Tr. 12126-27.) 

A number of arguments against the compatibility strategy 
were raised at RCA: 

First, if IBM customers could switch easily to RCA machines, 

then RCA customers could also switch easily to IBM machines. Beard 

5 'I 

• called this a "two edged sword". However, RCA felt that it "had more 

7 ; ! . . 

y * to gam . . . than to lose" because IBM had many more existing cus- 
^ ;, tomers than did RCA. (Beard, Tr. 8519-20; see also Rooney, Tr. 

.i 

9 I 11857.) 

LI | 

„:{ * Thus, the Spectra 45 was placed between the System/360 Model 40 

: j and the Model 50, and the Spectra enjoyed a price/ performance 

y I advantage over the 360 Model 40. However, the performance of the 

^ [ 360/50 was sup«rior to the Spectra 45. The same was true with 

1A J respect to the comparison between the Spectra 55 and the System/360 

— | Models 50 and 65. (3eard, Tr. 10106-07, see also Tr. 10097-99.) 

J The Spectra 4 5 and 55 were bid against the IBM System/360 Models 50 

^ j and 65. (Beard, Tr. 10113, 10121-22.) 

i 

■1 

\ 

\ -556- 



■V 

1 

i; 

i 

, i 

u 
£ i 



L 
2 

3. 

4. 

f 

5 

7 

3 

9 

10 

II 

12 

12 

14* 

15 

IS 

17 

IS 



Second, the similarity between Spectra and 360 "sharpened 
the comparisons" between RCA and IBM., making it "easier for the cus- 
tomer to analyze and quantify the differences" and putting "RCA in a 
position where its products could easily be criticized versus what 
IBM was offering. ... If there were any deficiencies on RCA's 
part they would probably stand out as weaknesses." (Beard, Tr. 8526.) 

Third, RCA could have chosen "the most natural alternative 
. . . an extension of the 301, 3301 systems". This would have pro- 
vided two advantages to RCA: it would have given it "a certain 
advantage" in marketing to the existing 3301 user base, because of the 
"software investment that [the users] had made in those machines". 
(Beard, Tr. 8524-26.)* And it might have enabled RCA to provide a 
"superior architecture to what IBM had chosen".** (Beard, Tr. 
8524-26.) 

RCA considered this important decision only briefly. 
McCollister testified that "because of the press of time in this case, 



;( * As noted above, RCA chose to forego this "advantage" and did 

19 .i not provide any emulation on the Spectra 70 series for the 3 301 

t user. 

I ** According to Beard this possibilitv did not seem to be very 

21 ! likely: 

•4 

22 :! "However, I think most of us felt that it really didn't 

| make that much difference to the customer what particular 

23 | machine instructions were made available; that the 360 
I set was a complete set, it included most of the things 

2- 1 ' we had thought of and perhaps some that we had not thought 

■! of; there were some things that were missing, but these 

25 I were secondary in our minds." (Tr. 8524-26.) 



-557- 



I am not even sure that there was a formal product proposal". (Tr. 
9630-32.) The compatibility arguments prevailed, and two or three 
weeks after the announcement of System/360, RCA decided to make its 
Spectra series "as compatible with the 360 as the circumstances per- 
mitted". (McCollister, Tr. 9273.) 

With the compatibility approach that RCA chose, its Spectra 
series had the same instruction set, instruction format and word 
length as 360 » However, "in terms of the engineering implementation 
of this architecture, it was quite different between RCA and IBM. 
.... If you took these machines apart, they were totally different 
machines . . . RCA used a completely different set of components . " 
(McCollister, Tr. 9644-45.) 

(ii) Commercial and Scientific Ability of Spectra 70 Series. 
As discussed above (see pp. 290-96), IBM's System/360 was aimed at 
all users regardless of application. Initially, RCA planned to market 
the Spectra systems for "commercial as distinct from scientific 
purposes . . . it' was a stated strategy to all of our marketing people 
that we were selling to the business environment and precisely said 
that we did not have a computer to compete in the scientific arena". 



:| -558- 

i 

;l 



L 
2 
3 

4- 



Series" was a "versatility for handling data processing, real time, 

and scientific applications from the small user to the very large". 

(DX 617, p. 2; see Beard, Tr. 9099-100.) Beard testified that 

the "primary reason" for making that a "salient feature of the design 

philosophy of the Spectra 70" was: 

"We felt that as the customer world became more sophis- 
ticated that there would be a consolidation in the 
computer type operations of more than one type of 
function and therefore this versatility, which allowed 
for engineering and scientific type problems, communi- 
cations problems, data processing, batch problems, 
information control systems . . . could be merged into 
one computer complex. 

"It may be a relatively small computer, if the 
operation is a small one. It could be a very large 
computer, if the operation had a large volume of data 
to be processed. There would be configurations where 
reliability was extremely important and it would be 
necessary to have multiple processors in order to have 
redundancy in the system. 



7; 

a; 

* j 

! 

10 : 

i 

t 

11 : 

12. | 
13 \ 

"It would be an advantage, from what we saw in the 
5- ;| field emerging, for the hardware as well as the software 

to accommodate these various functions in one system 
t - ii as opposed to having distinct unique systems for each 
* & of those functions." (Tr. 9100-01.) 



17 
IS 



Very soon after the initial delivery of the Spectra, the 

consolidation of the various types of functions which was anticipated 

•a* 

±* :j in "the design philosophy of the Spectra 70" had come to pass with 

;( 

2G ;j "some of the more advanced customers . . . ready for these types of 

.1 
-- ; ! systems in the latter half of the sixties, and certainly that trend 

22 1 has continued into the seventies". (Beard, Tr. 9101-02.) By 1970 

23 j RCA was advertising the versatility of the Spectra 70: 

r 

I 

2* i "The emergence of third generation equipment with 

j increased speed and storage capacity has brought us 
25 j to the realization that scientific applications are 



-559- 



L 
Z 
3. 

3 

a 

7 
8 
9- 
LO 
11 
12 
13. 
14- 
15 
16 
17 
13 
19 
20 
21 



within reach of almost every computer user. In the 
past these applications were confined to the big 
and expensive machines. 



"For all your, data problems — from simple account- 
ing to management science programs — Spectra 70 offers 
a complete systems approach. 

"Linear programming . . . statistical analysis 
. .. . simulation » . . automatic machine tool control 
. . . all are key elements of management science 
operations » Spectra 70 handles these applications 
and your normal data processing at the same time . " 
(DX 619, pp. 1, -39, emphasis in original.)* 

To assist its customers RCA offered its "Systems Scientific Services": 

"Systems Scientific Services provide a broad range 
of support, to RCA customers in software areas. 

"By supplying generalized scientific, statistical, 
simulation, and mathematical software, Systems Scientific 
Services assist the user in achieving efficient use of 
his system . . . right at the start. 

"Also available are scientifically oriented pro- 
gramming systems." ( Id. , p. B») 

(iii) Problems with Spectra 70 Series. The Spectra 70 series 

suffered from various problems that hurt its performance in the 

marketplace. Much of the equipment of the Spectra 70 series suffered 

from reliability problems which users took into account in choosing 



RCA also explained the new field of management science: 



• : i "Along with the advancement in equipment, the 

22 :j technology and management science has made significant 
,.| advances to the point that it is an integral part of 

23 j modern, efficient, organizational management. 



24 I "Management science has a broad definition that 
1 includes mathematical, statistical, and operation 

25 ; | research techniques that aid in effective decision 
making on the part of management." (DX 619, p. 1.) 

-560- 



z ' 

i 

3 

f 

7 
8 
9 

LO 
11 
11 

n. 

15 

16 



between RCA and IBM computers. (Rooney, Tr. 12190-91.) Rooney com- 
plained about RCA's equipment as late as June 1970: 

"RCA equipment apparently requires larger amounts of 
dedicated preventive maintenance time than that of our 
main competitor, IBM. Customers that have both our equip- 
ment and IBM equipment are aware of this , and this works 
to our detriment in the marketplace." (DX 621; see also 
Rooney, Tr. 12186, 12202-03.) 

Moreover, according to Rooney, RCA's equipment was 

"apparently more sensitive to environmental fluctuations 
than that of competition, particularly to IBM. This makes 
our customers somewhat sensitive to the differences between 
our maintenance policies and theirs. I am told, for 
example, the 360/30 's can be left without any maintenance 
whatsoever for weeks on end. Yet, most of our systems 
require that we take the system from the customer for 
periods of time every day." (DX 621; see also Rooney, 
Tr. 12145-48.) 

For example, RCA's disks were "more sensitive to air conditioning" 

than those of IBM, "so, if you did not have the adequate amount of 

air conditioning, that could lead to the need for more preventive 

maintenance". (Rooney, Tr. 12197-201.) 

RCA suffered problems during the installation of the 



*7 ,| Spectras. In that regard Rooney testified that: 



-S ;j "RCA equipment was more difficult to install because 

:| of certain environmental factors. I remember the RCA 

-- \ equipment required more air-conditioning and power and 

3 I remember a problem of size, physical size of the units 

-- ,| being involved, in terms of: if we replaced IBM, certain 

i of our units would require more physical floor size than 

21;! IBM equipment." (Rooney, Tr. 12175-76.) 

•i 

22 '\ These problems made it harder for RCA than for IBM to install its 

23 {equipment. (Rooney, Tr. 12204-05; DX 620.) It was reported to 
i 

24. 1 Rooney in 1970 that: 



-561- 



"In the area of installation, the RCA-IBM comparison 
is not restricted to just power and air-conditioning 
requirements. The problems are more profound, and 
bear directly on the equipment designed. 

"The installation of RCA data processing equipment 
has historically been more difficult and more time, con- 
suming than. that of our competition, particularly IBM's. 
Since the RCA marketing strategy is to sell to the IBM 
replacement market, the installation of RCA equipment 
is constantly being compared against IBM in an unfavor- 
able light." (DX 620; see Rooney, Tr. 12205-06.) 

During 1968 a portion of the marketing force was diverted 

from seeking new business to coping with problems of installation. 

At that time the marketing force was 

"very, very heavily occupied in working with existing 
customers on the installation of equipments which has 
[sic] been ordered at an earlier time. . . . [D]uring 
the year 1968 about 75 percent of [the time of] the 
marketing organization * . . was devoted to working 
with existing customers as opposed to seeking new 
business. . . . And this made very heavy demands upon 
^ ~| the time and capabilities of our field marketing 

organization, and this impacted to some degree our 
ability to get new orders." (McCollister, Tr. 9647-50, 
9653.) 



7 



\ 



i 



A 



RCA also found that its marketing force had to take time out from 
their normal selling efforts to deal with "[t]he problems of training 



* ij customers in the programming of the equipment, in working with the 

'I 

- :j customer in the installation of the equipment and the conversion of 

;{ ! 

- | his system of processing work to this new method". (McCollister, j 



- i Tr. 9649-50.)* 



3 -t * The amount of effort expended by the marketing force on customer ! 

j training was related to the fact that "the Spectra 70 equipment was 
& I new to the user" . There was a demand for the services of the mar- 

:| keting organization to deal with problems in systems programming for j 
15 j Spectra because there were new programming products and, as McCollister : 



-562- 



I 

2 

Z 
4 

5 
S i 

7 

3 

9 

10 

11 ! 
12 

13 



In addition to the problems that pervaded the entire Spectra 

line, RCA experienced problems which were uniquely associated with 

particular models. Those problems caused the Spectra product line to 

vary greatly in its degree of success . 

The Spectra 70/15 and 70/25. First deliveries of the 

smallest computers in the Spectra series, Spectra 15 and 25, were made 

toward the end of 1965. (McCollister, Tr. 9640.) The 70/15 and 70/25 

had less function than the rest of the Spectra line. This permitted 

them to be brought out earlier (Beard, Tr. 8460) : 

"They did not use integrated circuits . . . [i]t 
was a means of protecting us against any risks that 
there might be in the use of integrated circuits in 
the larger systems . 



"Secondly, we felt that there might be some cus- 
tomers who are interested in a system that did not 
have a complete instruction set and was simpler to 
14, ji operate." (McCollister, Tr. 9719.) 



The 70/15 and 70/25 were "relatively poor competitors" 

(Beard, Tr. 10110) and thus not very successful. (McCollister, Tr. 

9642.) The lack of a complete instruction set — one of the reasons why 

the 70/15 and 70/25 were introduced — and the limited capability of the 

systems were two of the liabilities of the 70/15 and 70/25: 

"It turned out that most customers wanted the 
systems which had the more complete capabilities, and 
also the 70/15 and 70/25 did not have the communica- 
tions capabilities that the larger systems had and they 
did not have the programming language capabilities that 



15 

16 ( 

17 \ 

IS j 
i 

19 ; 

20 : 
21 

22 

23 •, 

,\ put it, "the experience of the industry in general is that there 

Z-" ;) is always work to be done on new programming products". (Tr. 

9651-52.) 
25 



-563- 



r 

\ 

} 
1 

I i 

L 

a 

4- 



the larger systems had". (McCollister, Tr. 9719.) 

In addition, "there was no COBOL capability provided at all" on the 

70/15 and 70/25 (McCollister, Tr. 9730-31) , which was anomalous 

since the 70/15 and 70/25 processors "in general left out the 

scientific type of instructions, and concentrated primarily on the 

data processing instructions". (Beard, Tr. 9071-73.) 

In November of 1968 the product planning organization of 

the RCA computer division* made a similar observation: 

"[T]he Spectra 70/15 and 70/25, are basically sound 
processors, however, minimal software, no communication 
facilities , no random access hardware or. software 
facilities and the lack of slow speed/low cost card and 
print devices were the prime reasons for poor competi- 
tive position." (PX 127, p. 77.) 



The 70/15s and 70/25s were also hurt by RCA's absence of 
"marketing emphasis". (McCollister, Tr. 9729.) McCollister testified 
that the competitive position of these two systems "really wasn't that 
important to the RCA computer division". They were "insurance 

* i policies" using "existing technology that we could bring to market, 

. 

7 I deliver to customers before we could deliver the larger systems". 
•t 
i 

3 ;j (McCollister, Tr. 9740-41.) As a result RCA put little effort into 

« 1 

5 .} marketing the 70/15s and 70/25s: 

a! 



>7 I 

~ j ' * During the period 1964-1972 the RCA computer division had three 
*5 :! different names: from 1964 to 1968 it was called the "EDP Division"; 
"" i from 1968 to 1970 it was called the "Information Systems Division"; 
and after 1970 it was called the "Computer Systems Division". Here 
it will generally be referred to as the "computer division" or 
"Computer Systems Division" unless appearing in a quotation. 



La 



je 



-564- 



L ! 
2 
3 i 

5" i 
» 

i 

T I 

3 

9- 

a 

12 
13 

14* 

15" 

^ , 

* In distinct contrast to the RCA lassitude with the low end of 
17 I its line, IBM constantly attempted to grow the market with its low 

; end computers. Thus, the 1401 (announced in 1959) and the 360/20 
IS | (announced in 1964) were the largest-selling IBM systems of their 

j time. (See above, pp. 141-42, 399.) 

19 i 

j ** The Spectra 45 was first installed in July or August 1966 and 

20 ! the Spectra 55 first installed toward the end of that year. Deliveries 
i in quantity of the 70/45 began in 1967. (McCollister, Tr. 9640-42.) 

21 :■! 

i /Beard wrote in 1965 (and testified as to the accuracy of his 

22 1 statement) that "the Spectra 70/45 is a medium-scale processor with 
a high performance capability for business, scientific, communications, 
and real-time applications", giving airline reservations or brokerage 

| quotations as forms of real-time applications. (Beard, Tr. 9080-81; 
24. 1 DX 617, p. 7.) 
I 

I 

i 

•! -565- 



"These were relatively low cost, low margin systems, 
and when we had a finite amount of marketing field 
manpower, it made better economic sense for us to 
concentrate on the larger systems , where we had larger 
unit sales value J* (McCollister, Tr. 9724-25.)* 

The shipments of 70/15s and 70/25s turned out to "trivial". 

(McCollister, Tr. 11355.) And RCA produced the 70/15s and 70/25s 

"only during part of the total life cycle of the Spectra 70 family". 

( Id. ) 

The Spectra 70/35, 70/45 and 70/55. The Spectra 70/35, 

70/45 and. 70/55 were larger processors than the 70/15 and 70/25. 

Deliveries of those systems were about "fifteen months or so behind 

IBM".** (McCollister, Tr. 9646.) The 70/45 turned out to be the 

"most successful" of RCA's Spectra series./ (McCollister, Tr. 

9665.) The 70/55 was less successful than the 70/45. It suffered 

from several problems: 



(1) The 70/55 "had serious memory problems. . . 
We would get repeated errors in memory due to tech- 
nical failure in the memory itself and this would 
bring the system down. [We] had a great deal of 
difficulty in maintaining the gear and keeping it 
up." (Rooney, Tr. 12139.) In fact, "there was some 
exchange of memories . . . some early number of the 
first machines had to have their memories replaced." 
(Beard, Tr. 10111-12.) 

(2) RCA experienced "manufacturing problems 
with the [70/] 55s, which gave us an unusual amount of 
field maintenance attention during the first year" . 
(Beard, Tr. 10112.) 

(3) The 70/55 was difficult to install and 
relocate. (DX 620, pp. 1, 3-4.) 

(4) The 70/55 came out approximately a year 
after the 70/45 and tended to be "eclipsed" by the 
70/60 and 70/61 which RGA brought out shortly there- 
after. (Beard, Tr. 10109-10.) 

2Q '■[ (5) The 70/55 was hurt because it did not offer 

2i ;j any emulation capability. Notwithstanding the fact 

22 ' : j that emulation capability was important to the success 

23 ij of the 70/45 system, the Spectra 55 did not emulate 

24 !J anything. (Beard, Tr. 10109, 10233.) 

25 ,i 



-566- 



L 
2 

3 

4* 

5 
a 
7 
3 
9 
10 

12 
13 

14- 

15 



The result of these failures of the 70/55 had a "dampening 
effect on the [RCA] sales force" and led to customer cancellations. 
(Beard, Tr. 10111-12; see McCollister, Tr. 11216.) 

While the 70/35 provided for emulation of the 301, that 
emulation "did not work successfully because the Spectra 35 was priced 
at such a high price that it was not a logical move for the 301 user 
to move up to the Spectra 35 system. . . . [Thus,] 301 users did not 
move up to the Spectra 3.5". (Rooney, Tr. 12137-39.) 

The Spectra 70/46, 70/60 and 70/61. RCA's 1969 Annual 

Report described the 70/60 as a 

" [l]arge-scale . . . batch processor, which is designed 
to handle retail credit and reservation systems, auto- 
mate production control, and service government and 
industry data banks.'* (DX 674, p. 8.) 

The 70/46 and 70/61 were time-sharing systems. RCA began its work on 

time sharing during 1967 by attempting with the 70/46 "an expansion of 

the 70/45". (McCollister, Tr. 9673-74; DX 672, p. 21.) McCollister 



la i testified that the hardware for the 70/46 was "in its elements iden- 



17 
IS 



tical" with that of the 70/45 with "the addition of some faster 
registers in the machine". He estimated the hardware development 



19 : i effort of the 70/46 was "in the order of $2 million . . . because we 
2Q ] made so much use of what was already existing in the 70/45". 

21 ; (McCollister, Tr. 9679.) The Spectra 70/61 had a comparable 

22 1 relationship to the 70/60 in terms of design approach as 

i 

<n ! the 70/46 had to the 70/45. (McCollister, Tr. 9680.) 

~ i 

74, | RCA, like GE and IBM (see above, pp. 417-18, 505-06), 

i 
2c 1 went into computer time sharing because of the changing demands of the 



-567- 



r- !: 



industry. The introduction of the 70/61 "was in response to accel- 
erating industry shift from traditional batch orocessina to r^nte 
computing, a system in which a large central computer accepts and 
^ jL almost simultaneously feeds back data to numerous remote terminals" . 
(DX 674, p. 8.) 

As a result of "[t]he growing acceptance of remote com- 
puting", RCA foresaw "excellent potential for sales of data communi- 
cations terminals and other peripheral equipment as well as for com- 
puter hardware" and expanded its manufacturing capabilities for 
peripheral computer equipment. ( Id. ) 

However, orders for the 70/46 and 70/61 during the years 
1969-1971 were "less than had been projected in the forecasts because 
. . . the 70/46 had originally been planned as a system from which to 
gain experience with this class of product". (McCollister, Tr. 9694.) 
The marketing forecast was "excessively optimistic". (McCollister, 
Tr. 9695.) 

RCA, like GE and IBM, ran into substantial difficulties 
developing the time-sharing software. RCA's time-sharing software was 



r 
i 

T 
3 
3 

a 

2 
3 

A 



.s 

LS 



•J called the Time Sharing Operating System (TSOS) .* The development of 
. TSOS was "[b]y far the largest software development or largest pro- 
.} gramming system" that RCA's computer division had undertaken. 

21 '\ (McCollister, Tr. 9697.) 

22 : ; 



^ | * At a later time TSOS was known as VMOS . (McCollister, Tr. 
| 97X7-18.) 



-568- 



1 
z 

3 
4> 
5 
S 
7 
3 
9 

ia 
ii 

12! 

13 

IA! 

15 

16 

17 

IS 

20 ' 
21 



Despite this effort RCA had 

"[i]mportant difficulties with . . . TSOS-VMOS which 
substantially impaired, certainly in early installa- 
tions, the performance of the system as a whole, which 
includes both hardware and software." (McCollister, 
Tr. 9710-11.) 

The problem with TSOS was that "there were bugs in it, which took time 

to get rid of, and it was late, as far as providing functions 

specified were concerned". (McCollister, Tr. 9694.)* Similarly, 

Rooney, who joined RCA in 1969 and was President of the Computer 

Systems Division in 1971 and 1972, testified that: 

"The Time Sharing Operating System was a form of 
virtual memory system that had a great deal of func- 
tional capability to offer, that was new and unique ■ 
in the marketplace, but its reliability in performance 
was extremely poor and we had not achieved a high 
degree of reliability with that system while I was at 
RCA. 



"The system was referred to as bombed out'. There 
would be a problem. It would essentially go down. 
There had been a malfunction in the hardware, but in 
essence it was what was referred to as a bug in the 
program of the operating system, but it was not able 
to cope with handling certain data, as it was speci- 
fied to handle it." (Rooney, Tr. 12132-34.) 



* 



McCollister observed that developing time-sharing software was 
ii difficult "for other manufacturers attempting [it]" as well as for 
ij RCA. (McCollister, Tr. 9694.) He added: 



2- ij "The history of the computer .industry is filled with 

^ ,! examples of difficulty that every manufacturer has had 
with the introduction and initial installation of new 
products, either hardware or software. This is a normal 
way of life. 



23 
24 
25 



" [S]oftware, in particular, is a complex technical 



-569- 



L \ As a result 

Z 

3 



4 
5 

6 ; 
7 
8 
Si 
10 

11 \ 
12- J 
13 [ 

i 
i 

1± ! 
** i 

I 

15 ] 



"It was available to the user, but there were a 
great many periods of down time and, also, if you 
were operating with terminals, the response of the 
system would be very s«low in a timesharing mode." 
(Rooney, Tr. 12134.) 



In July 1971 r Rooney identified "inadequate software" as among the 
"functional capabilities which detracted from [Spectra's] ability to 
meet its product. objectives" . (DX 11101, p. 1.)* While the per- 
formance of TSOS/VMOS improved "as time went on" (McCollister , Tr. 
9718) , development work on TSOS continued until RCA left the computer 
business. (McCollister, Tr. 9674.) 



task, and very large programming systems have literally 
hundreds of thousands of instructions in them between 
which there is interaction, and sometimes you don't 
know whether or not there is a fault until for the 
first time a particular combination occurs, and it is 
very difficult to develop this total mass of logic and 
to make certain that it is all correct at the time you 
turn it over to a user of equipment. This is a common 
^g -J experience within the computer industry. 



17 !| 

'I 

18 i 

19 ;« 
20 



"For another reason, competitive conditions in the 
industry have been such that the manufacturers are 
always under extreme pressure to get out a new set of 
equipment and get it into the marketplace and get it 
installed as quickly as they can." (McCollister, 
Tr. 9696-97.) 



i * We are aware that this document is not in evidence; however, 
-^ :', we believe it is reliable and rely on it because it was written 

■i by a person with knowledge of the facts (J. W. Rooney, President 
22 J of RCA's Computer Systems Division) contemporaneously with the 
^ ;j events described and confirms Mr. Rooney ' s trial testimony on 
-« i the same subject. (See Tr. 12131-37.) 

24 ;i 

25 i 

i| -570- 



i 

2 

3 

4- 

5 
6 
7 
3 

io i 

i 

ii i 

12 

13 i 

! 

t 
i 

15 ! 

i 

I 
15 I 

17 I 
13 | 

19 i 

I 
21 ii 



Other Product Problems. In addition to problems with its 
processing units, RCA experienced problems with the operating software 
(in addition to time-sharing software) , random access memory units, 
card readers and memory stacks on Spectra 70 . These problems created 
substantial problems for RCA in its marketing of Spectra. As Beard 
testified, in addition to looking at the price/performance or through- 
put claims of RCA, a user to determine the value of a computer system 

"would be looking primarily at the total system service 
that he would be supplied with, not only the effectiveness 
of the hardware to perform his requirements, but what soft- 
ware would be made available, and how effective the software 
was, the caliber of the maintenance organization to maintain 
the equipment once it was installed, his impression as to 
how well he would be supported by the vendor on future 
applications which he had not yet fully defined in his own 
mind in terms of new equipments, new software." (Tr. 
10090.) 

In addition, the user would take into account the relative functional 

capabilities of the peripheral equipment offered by competing systems. 

( Id. ) When compared to IBM, the overall "value" of Spectra to the 

user did not match IBM's 360. 

TDOS. For its Spectra systems from the Spectra 45 up, RCA 

I had an operating system called TDOS. Rooney testified: 



"When it was announced it was a good system, but 
RCA did not continue to improve upon it at the same 
pace as IBM improved upon their OS. Our system, while 
performing satisfactorily in terms of reliability, did 
call for a lot more operator intervention in terms of 
22 ';! performing the work than the OS system. 

i 
22 i "1 made a strong plea for an improved system 

j called OS 70, which was under development, to be 
ii used on the RCA 6. That system was decommitted in the 
z ' ! | early part of 19 71. 

25 ! 



-571- 



"By ' decommitted ' I mean it was never brought to 
the market. 



"The people responsible for putting the system 
together felt that they couldn't do it in the time 
frame that had been asked for". (Rooney, Tr. 
12135-37.) 

Peripherals and Memory Stacks. As discussed, RCA intro- 
duced its RACE file during 1964, and it suffered from various 
problems. (See above, pp. 549-50.) RCA marketed the RACE with its 
Spectra series and continued to market it actively into 1968 at 
which time it 

"was impacted by the progressive development of disk 
file technology. Disk files were more reliable 
devices. There were fewer things to make mechanical 
trouble in them. They had a faster access time and a 
faster transfer rate of information from the medium into 
the processor, and as the cost performance characteris- 
tics of disk files improved, the relative advantage 
and cost performance of the so-called RACE unit was 
reduced, until you reached the point where, for most 
applications, a disk file, as illustrated by the 2314, 
was a preferred approach." (McCollister, Tr. 9659-60.) 

During the middle 1960s RCA still was not producing disk 

drives. To meet the demand from users for disk drives, RCA purchased 

; IBM 2311 and then CDC 2311-type disk drives for use with the early 
deliveries of the Spectra 45 and 55 in 1967. (Beard, Tr. 9935.) RCA 

i did not deliver its own 2311-type disk drive until the end of 19 6 7 or 



f ! 

" :j beginning of 1968, a year and a half after its first Spectra deliv- 

1 ! * . 
1 eries. (Beard, Tr. 9913.) 

When IBM began deliveries of its 2314 disk drive, RCA found 

4 | that its marketing people "were under a handicap in selling the Spectra 
e i 

- :! 
I 

■I -572- 






70 Systems. We did not have a comparable product to the IBM 2314 at 
the time." (Beard, Tr. 8575.) 

RCA's development of a 2314 equivalent was hampered by the 
departure of the group of persons who had worked on development of 
its 2311-type disk drive in 1967-68 to form another company, which 
g | was called Linnell Electronics.* The departure of those persons 
7 j| impaired RCA's ability to develop new disk drives. (Beard, Tr. 
g 'it 9924-28.) During. 1968 RCA determined, that Memorex was ahead of RCA 
in disk drive technology, and RCA contracted to have Memorex supply 
RCA with its "first year or year and a half supply of disks". Obtain- 
ing disk drives from Memorex "cost additional money" because RCA 
"had in parallel [its] own development going on which was going to 
be about a year later than Memorex 1 s". However, RCA could not 
afford to wait the additional year because it was "losing too many 
sales" to IBM "for the lack of it". (Beard, Tr. 8574-75.) 

RCA supplied its own controller for the Memorex 2314-type 
drives (Beard, Tr. 10254-55) , developing it at an engineering cost of 
: | about $500,000. (Beard, Tr. 10246.) RCA started to work on a 2314- 
:| type product as a "full design project" in 1968 (Beard, Tr. 9922) and 

13 :; 

j delivered the first units to customers "around the latter part of 1969 

20 ': 

! to perhaps mid-1970". (Beard, Tr. 9915.) The fact that RCA "was not 

21 ] ■ 

:i able to produce on its own or to duplicate the 2311 or the 2314 disk 
*2. » 



9> 
LC 
U 
12 
12 
14- 
13 



16 II 



17 



■\ 



" i 

j * In 1968 Linnell was manufacturing IBM-compatible 2311-type disk 

2- ) drives for use with System/360. (DX 12543.) By 1972 it was manufac- 

;! taring 2314-type compatible disk drives as well. Both drives were 

25 i marketed by Bryant Computer Products. (DX 4556, p. 2.) 

i 

■t 

"\ -573- 



drives until very much after the IBM delivery" hurt RCA. (Rooney, Tr. 
12123, 12192-94.) Disk drives offered "a functional capability very 
much needed in terms of price/performance in the competitive market- 
' place and without that capability you were in a weak competitive 
situation against IBM" ► (Rooney , Tr. 12192-94.) Thus, in July 1971 
I |j Rooney wrote that "the lack of a 2314 competitive device until late in 
the product life" was one of the "many functional capabilities which 
detracted from [Spectra's] ability to meet its product, objectives" . 
(DX 11101, p. 1.) 

RCA also had initial difficulties with the card reader on 
its Spectra series ► The difficulties were corrected only at the cost 
of making it a "very high cost product". (McCollister, Tr. 9604.) 

One of the most severe problems faced by RCA was that 
of providing reliable memory stacks in 1967 and 1968. This problem 
cost RCA as much as $10 million and caused J. R. Bradburn, Executive 
Vice President of the computer division, to write to R. W. Sarnoff, 
President and Chief Executive Officer of RCA (PX 338, p. 49) in 
December 1968 recommending that the Memory Products Division be 
transferred to his division: 



I 

I 

L 
1 i 

3 I 

L 

5 

5 



:{ 
i 



1 "Development processes must involve more than 

>A 1 theoretical analysis and its immediate physical 

; ! embodiment. A thorough understanding and consideration 



'C ! 



"Modern computing and data processing systems j 

£ ; | consist in essence of input/output peripheral equip- j 

i ment, control, and memory. Development of complete ; 

11 :] competitive systems involve [sic] simultaneous, j 

j continuous, and coordinated development of all com- ; 

2. ;| ponents. The single most important element of this 

j overall development is memory. ' 



-574- 



L 
2 
3 

4. 
§ 



of mechanical design., reliability, manufacturability , 
and maintainability of a complete memory system is 
required. Nothing less can meet competition today. 



"The present organizational structure within RCA 
is not conducive to efficient operation or to meeting 
these requirements. It does not bring to bear upon 
the decision making, process the needed emphasis or the 
proper sense of order of importance adequate to meet 
£ |t the needs both in the short and long runs . 



3 

La 
u 

12 

13 
14- 
15 
IS 



" . . * [This] is what has been demonstrated by 
the inordinate difficulties 'encountered in trying to 
provide reliable memory stacks for our computer ship- 
ments in 1967 and 1968 - Poor stacks may have cost us 
as much as $10,000,000 in those two years. Additionally, 
our problem is portrayed by what has been inadequate 
provisioning in the engineering budgets of Memory 
Products." (DX 840, p. 1.) 

(iv) RCA Success with the Spectra 70 Series. Despite the 

numerous problems experienced by the Spectra 70 series, during the 

period of its life (1965-1969) RCA enjoyed considerable success with 

its computer business. McCollister testified that: 



"1966, "67, '68 and r 69 were generally periods of 
steady and encouraging growth, and we were operating 
for the most part during this period at pretty much a 
■| breakeven, although in the year 1966, because we had 
W it a heavy installation workload, we upped the budget of 

'.I the marketing organization, which threw us back into 
-S :j the red, and in the following year or two we began to 

increase the amount of money going into the engineering 
organization . 



19 



20 ;j "But I would say that beginning in 196 5 through the 
i year 1969 we were making what appeared to be encouraging 

21 ; ! - progress. We did have an ability to compete within our 
i particular scope of operations and the corporation was 

22 1 encouraged about the long term outlook for the Division." 
(Tr. 9246-48.) 



23 ;| 



24> 
25 



I RCA's annual reports confirm McCollister ' s assessment: 

1 

■J In 196 4 RCA reported that its "gross computer sales and rentals" were 



-575- 



higher than $100 million, having grown from $14.6 million in 1960. 
"RCA's total data processing business earned a profit for the full 
year." (DX 669, p. 2.) 

In 1965 RCA reported that its profits in the computer busi- 
ness continued and "the potential for future profits was enhanced by 
the booking in 1965 of orders for 92 percent more computer systems 
than in the preceding year. By 1970, profits from the data processing 
business . . . are expected to become a highly significant factor in 
RCA's total earnings." (DX 670, p. 7.) 

In 1966 RCA reported that "domestic orders for RCA computers 
and their associated equipment rose by 53 percent over the 1965 level". 
During 1966, with Spectra deliveries beginning, RCA enlarged its field 
marketing force by 45 percent, planned for another 35 percent increase 
in 1967, and "boosted production capacity by 75 per cent to fulfill 
. the growing demand for Spectra 70 computers and other data processing 



. It equipment". (DX 671, pp. 4, 15.) RCA reported that a loss in its 
'] computer business in 1966 was caused by an increase in leasing as 
] opposed to purchasing by customers.* ( Id. , p. 4.) 

3 i 

:| In its 1967 Annual Report RCA reported that: 

ff i 

:l * Concerning the change in favor of leasing by customers RCA 
- ! reported that: 

4 
2 I "The increase in lease transactions was common 

■j throughout the computer industry and reflected in part 
2 j the tightness of the money market. While tending to 
promote long-range stability by spreading income over 
the period of lease, it reduced immediate income. This 
trend, as well as increased spending for future growth, 



)A 






contributed to a loss in RCA's computer business ror 



i 1966." 

•t 

} -576- 



Li 
2 j 

I 

i 

si 

» 

i 

7 I 

t 

si 

i 
i 

*! 

10 

i 

111 
i 
i 

12 | 

13! 

14 

15 

16 

17 

13 

19 

20 

21 

22 

23 

24 

25 



"Our computer business continues to grow at a faster rate than 
that of the industry as a whole. However, because most of our 
equipment is leased, rather than sold outright, income is neces- 
sarily delayed. In 1967, this situation placed us once more in 
a loss position, but we look upon this as an investment in future 
profits and we look forward with confidence to the period when 
our data processing activity will become one of the most importair 
parts of our business, surpassing even color television." 
(PX 338, p. 5.) 

In 1969 RCA reported: 

"Domestic bookings of RCA computers were more than 40 per cent 
greater than in 1968 and represented nearly a threefold growth 
in the past five years. 



"As the decade ended, our backlog of computer orders was 
30 percent higher than a year ago. The trend toward remote 
computing systems as well as the mushrooming of data processing 
applications make the outlook for the '70s very promising. We 
expect the dollar value of our information system shipments to 
increase significantly during the 1970' s and to approach $1 bil- 
lion annually. 



"While total revenue from sale and lease of computers 
rose 23 per cent during 1969, RCA's computer operations remained 
in a loss position. This deficit is largely the result of 
expenditures aimed at future growth, which include the building 
up of our marketing forces and expansion of software and other 
aspects of the business." (DX 674, pp. 8-9.) 

During the period 1965-1969, RCA's U.S. EDP revenue rose 

from about $89 million to approximately $211 million. (DX 8224, p. 2. 



RCA added 



"A pattern of fluctuating profits and occasional losses is to be 
expected in the development of a strong base for the future in 
this complex and competitive field." (DX 671, p. 4.) 



-577- 



c# RCA Computer Systems Division 1969-1971. The story 
of RCA's involvement in computer s at the end of the 1960s into the 
early 1970s has three parts: 

First t the change in personnel at the corporate level and 
in the computer division during that time frame and resulting 
changes in goals of the computer business; 

Second , the decision to develop and the consequences of 
the RCA Series ; and 

Third, the problems that resulted from the preceding two 
decisions. 

( i ) Changes in Management Personnel and Goals . On 
January 1, 1968, Robert Sarnoff became Chief Executive Officer of 
RCA while continuing as President; his father, David Sarnoff, 
continued as Chairman of the Board. (PX 338, p. 49.) During 1963 
Chase Morsey, Jr. left Ford and joined RCA as Vice President of 
Marketing, and the next year became an Executive Vice President of 
RCA. (McCollister, Tr. 11156-57; PX 338, p. 49; PX 339, p. 49.) In 
1970 Robert Sarnoff replaced his father as Chairman of the Board 



\ (DX 674, p. 36), and in 1971 A. L. Conrad became President and 
'; Chief Operating Officer. (PX 341, p. 38.) Conrad had worked his 



T i 



way up through the RCA Service Company (PX 400) which, " [i]n addition 



: 4 to its work for the government and in education, the Service Company 

| install [ed] and maintain [ed] home-entertainment products, commercial 
3 ; 



-578- 



I.: electronic systems, and business and industrial equipment". (DX 

2 : 677, p. 13.)* 

3 | At the time Robert Sarnoff became RCA's Chief Executive 
4»|[ Officer, RCA underwent a change in its corporate philosophy. In its 

1968 Annual Report RCA reported that: 



3 

a 

7 

3 

9 

10 

II 

12 

13 

15 



"In its formative years, RCA's growth depended primarily 
on a single product or service. 



"The word that best characterizes the modern RCA is 
diversity. An almost even balance has been achieved between 
manufacturing and service operations, and the well-being of 
your company no longer hinges upon any single activity. 
Carefully planned diversification has moved us into new 
areas of enterprise, such as vehicle rentals through The 
Hertz Corporation and book publishing through Random House. 
New businesses have been created from within, such as the 
Information Systems Division". (PX 339, p. 11.) 



RCA's desire for diversity caused it to acquire many 
different and unrelated businesses. Starting in 1966, RCA acquired 
the Hertz Corporation (automobile rentals) ; Random House and Ballan- 
^ :! tine Books (book publishing) ; Coronet Industries (carpets) ; Banquet 
17 ; Foods (frozen foods) ; Oriel Foods and Morris & David Jones (wholesale 
IS ; food distributors in the United Kingdom) ; Cushman & Wakefield (real 
13 1 



20 i * RCA's changes in higher management have continued to the 

j present. On December 31, 1975 Sarnoff resigned as Chairman of the 

21 ! Board of RCA. (DX 951, p. 36.) He was replaced as Chairman during 
j 1976 by Conrad who remained as President. (DX 13852, p. 37.) In 

22 :i September 1976, Conrad resigned as both President and Chairman, and 
Edgar Griffiths replaced him as President, while the position of 
Chairman remained vacant. ( Id. ) On January 1, 1980 Griffiths 

! filled the vacant position of Chairman and Maurice Valente became 
24. j President. (DX 13902. p. 48.) In June 1930 Valente was forced to 

i resign as President, and RCA abolished the position and created an 
25 j Office of the Chairman consisting of Mr. Griffiths and five other 

f j current executives. (DX 13861.) 

( 

;| -579- 

i 
i 

.i 



estate) ; Alaska Communications System (communications) ; and a color 
tube manufacturer in the United Kingdom. (Conrad, Tr. 14002-05; 
DX 671, p. 5; DX 674, p. 21; DX 677, pp. 14-19.)* 

During that period it entered several new businesses 
including the domestic common carrier satellite business. It also 
undertook, significant new investments in Colortrak (an advanced 
color television receiver) , SelectaVision, VideoDisc, Global Communi- 
cations and the RCA Service Company (related to lease and maintenance 
services for private telephone systems). (Conrad, Tr. 14002-05.) 

Thus, by 1971 RCA was a conglomerate engaged in a large 
variety of different fields including home appliances, televisions, 
radios, recording devices, federal defense contracts, communications 
services, broadcasting, automobile rentals, food, carpets, books, 
records, and real estate — and computers. (Rooney, Tr. 12022-24.) 

During the 1966-76 period, the chief executive officers 
!j of various RCA divisions and subsidiaries, including Banquet Foods, 
Coronet, Hertz, Random House and Global Communications, were directors 



i i 

i 

l ; 

i 

■L 



7 ; 

s] 

.1 



of the RCA Corporation. No officer of the computer division was a 

member of the RCA Board of Directors during this period. (Conrad, 

| Tr. 14027-28; PX 339, p. 49.) In 1971 revenues from the operations 



■Q 

| of the RCA Computer Division represented less than 10% of RCA's 
II : | 

?2. ■•; 



22 ■ * Bv 1979 RCA had sold Alaska Communications System and Cushman 
] and Wakefield. (DX 951, p. 17; DX 13902, p. 14.) As of 1980, RCA 
Z*. j had sold Random House and was trying to sell Banquet Foods. (DX 

■J 13860.) 



-580- 



J- i corporate revenues — $270 million out of $3 billion. (Rooney, Tr. 
12025.) 

At this time changes were also occurring in the personnel 
and goals of RCA' s computer division. During 1969 and 1970 RCA 
hired people from IBM to manage parts of the computer division. For 
example, L. Edwin Donegan, Jr., became Vice President of Sales in 
1969 and General Manager in 1970. (McCollister, Tr.. 11590.) 
Joseph W. Rooney came from IBM in 1969 and after a brief corporate 
staff job became Vice President of Marketing; in 1971 he was 
President of the Computer Systems Division. (Rooney, Tr. 11687-88.) 
V. Orville Wright was hired in 1970 and the next year was head of 



2 

3 

4- 
5 
S 
7 
8 
9 

La 
u 



£2 i! Systems Development. (Wright, Tr. 12785.) Sam Adams was responsible 



13 

14- ! 

16 



for business planning. Bill Acker was put in charge of the financial 

operation. (McCollister, Tr. 11590.) 

The goals of the computer division changed with the new 

corporate and computer division management during this time. Until 

.- II Robert Sarnoff took over in 1968, the RCA computer division had 

TS j placed its emphasis on accomplishing its business plans and obtaining 

ii 
ta 1 moderate growth, an emphasis which McCollister considered "correct". 

_ \ (Tr. 11156-57.)* 

20 i 
,1 



22j * Withington similarly identified and endorsed RCA's emphasis 
1 during that time : 



23 I 



"In 1965, the RCA strategy was to offer a line of general 

24 | purpose computer systems . . . with instruction set com- 
j patibility to IBM processors, . . . and to grow at a 

25 :; modest rate commensurate with maintaining profitability 
at all times. 



-581- 



In 1968 RCA changed its emphasis to one of quickly obtaining a 

larger market share. McCollister attributed this change to the 

newly-arrived Chase Horsey,* and testified that the change was not 

"beneficial" to the computer division because: 

"It tended to place the emphasis upon increasing market 
share and relatively deemphasize control of expenses and 
achieving a profit, and the end result is that the expenses 
in the RCA Computer Division mounted to the point where 
they contributed significantly to RCA's withdrawal from the 
business. 

"In other words, you place the emphasis upon share of 
market and you tend to deemphasize some of the other important 
aspects of running a successful business, and share of market 
is only one consideration". (Tr. 11158.) 

In its 1970 Annual Report RCA reported that: 

"Our highest priorities today are the establishment of 
a profitable computer business and the capture of the 
domestic industry's No. 2 position. RCA has made a greater 
investment in this effort than in any prior venture in its 
history, and we are convinced that the returns will be sub- 
stantial". (PX 340, p. 3.) 

This change in strategy was visible outside RCA. Withington 

> ;! testified that: 

r •{ 

'I , — 



"That was RCA's historic strategy since the announcement of 
the Spectra 70 series, . . . the system seemed versatile and 
attractive, and I saw no reason why RCA could not continue 
growing at a modest rate and increase its total market and 
perhaps even its market share on the basis of its current 
offering". (Tr. 56702-03.) 

* McCollister believed that, having come from the Ford Motor 
Company, Morsey "was very conscious of share of market statistics 
and he was also influential, and this caused the division to give 
increasing recognition to share of market as such". McCollister 
regarded Morsey 's emphasis on share figures as "exaggerated" and 
"a legacy from his experience in the automobile industry". (Tr. 
11156-58.) 



-582- 



"... RCA's new strategy involved attempting to grow very 
sharply in market share from the position they were in by 
means of attempting to capture IBM 1 s customers in the processes 
of evolving from the 360 series to the 370 series . . . RCA's 
new strategy involved expanding its operation in every respect, [* 
field organization, the manufacturing capability, and the 
engineering force, at a substantial increase in cost; and more 
importantly, it involved the anticipation of incurring the 
manufacturing costs of a very large number for RCA of new 
systems in a short time. . . . These financial requirements 
would inevitably cause a lack of profitability for a period". 
(Tr. 56707-09.) 

It was this change in strategy that led directly to the 
ill-fated RCA Series. 

(ii) The RCA Series . During the 1968 time frame RCA 

realized that if it was to achieve its new growth objectives it 

would need successor systems to the Spectra series. Unfortunately, 

RCA was undecided about what successor products to develop. 

McCollister testified: 

"In looking at the next family or generation of equipment 
beyond Spectra, there was a lengthy debate between the people 
responsible for programming systems, that is, the so-called 
software organization, and the people responsible for hardware 
or equipment specifications , and perhaps the engineering 
organization as well, as to exactly what the nature of this 
product should be." (Tr. 9809.) 



19 ! 

'» line. The first attempt, referred to at trial as the X series, was 

decommitted in 1969 for two reasons: first, RCA felt that it could 



20 ;. 



21 
22 ! 
23 
24 
25 



* For example, during 1971 the computer division expanded its 
operations: it opened a sales office in the United Kingdom to 
market RCA computer products in Europe (Rooney, Tr. 12365) and it 
constructed a new manufacturing site in Marlboro, Massachusetts. 
(McCollister, Tr. 10963-65.) 



-583- 







not meet what it predicted to be IBM's announcement of 370*; second, 
the series included 

" Cain architectural problem [in] that they doubted they would 
ever be able to complete the product line without a major 
restructuring of their whole development program. ... My 
understanding at the time was that they could not build it at 
ail if they had developed it or had set up the architecture". 
(Rooney, Tr. 122^5-26.) 

The second attempt at a successor to the Spectra was the 
New Technology System (NTS) . The NTS was originally scheduled for 
announcement in early 1971. However, " [tjhere was a slippage in 
that program and it was subsequently put off for announcement for 
approximately 18 months as a result of development problems within 
RCA itself". (Wright, Tr. 13173.) 

RCA made only "marginally small" investments in NTS. 
(Withington, Tr. 57079.)** It appeared that NTS, if announced as it 
was being developed, would have encountered competitive difficulties. 
Withington testified: 



"The basic reason for my concern was that I believed at 
the time that IBM would introduce a new family of general 
j !( purpose computer systems in the time frame 1973 to 1977, which 
It was the time frame in which RCA's NTS computer systems were to 
- J be shipped. 



9 



* This prediction was based on "intelligence" from people who 
£ I had worked at IBM and from trade journals. Based on this intelligence 

i RCA understood that IBM announcements would begin in the second or 
n | third quarter of 1970, with larger models announced first, and 

- : ; announcements continuing throughout 1970 until the lower models of 
•! the line would be announced at the beginning of the 1971 period. 

— } Shipments were anticipated to be 12 months later. (Rooney, Tr. 
12225-29.) 



.2 
15 



** During this time RCA hired A.D. Little to review RCA's product 
and marketing strategies in its computer business. (Rooney, Tr. 
11814.) On behalf of A.D. Little, Withington made suggestions and 
wrote a report. (DX 2666.) 



-584- 



^ t 

i 

z\ 

4* 

5 

6 

7 1 

3 

9- 
Id 
II 

12 

13 
14 
15 
IS i 

17 | 
IS I 

19 i 

> 

ZQ j 
21 I 



"I believed that the nature and functionality of the NTS 

line would be inadequate to meet the needs of customers who were 

IBM users or who would otherwise consider IBM systems during that 
time frame."*- (Tr. 56715-16.) 

Withington told RCA that "a revision and acceleration of 

the product plan would be necessary if RCA would have . ► . ' [a] 

good chance of attaining the desired market share'".** (Tr.. 56716; 



* Withington listed the ways in which the NTS line would 
be "inadequate in. terms of functionality": 

"I noted that RCA's NTS line involved only mono- 
processors rather than multiprocessors. I noted there 
would be no small satellite processors, and I noted 
that NTS was to be aimed at the medium monoprocessor market 
segment, this meaning in terms of price range, and noted my 
belief that the IBM customers would be migrating from the 
medium monoprocessor market segment to large multiprocessors 
to obtain the advantages of centralization. 



"I foresaw that the IBM systems of the 1973 to 1977 range 
would be equipped with multiprocessing, satellite systems, 
virtual memory, and fail-soft, of which all but the latter 
were in fact announced; and that there would be a trend toward 
centralization, which did take place. 

"Since RCA's NTS line did not fit with either that 
functionality or that trend, I believe that was the entire 
reason for my concern". (Tr. 56715-16.) 



** He recommended that RCA embark on a development program for a 
satellite processor and on a "comprehensive terminal plan" because 
"an adequate line of inquiry interactive and remote batch terminals 
must be offered with network systems": because of their "visibility" 
and large numbers, "terminals may dominate users' selections". The 
H existence of a terminal product line for RCA "might very well" have 

22 | made a difference in the success of the NTS plan because "the 

j profitability of the computer business as a whole might have been 

23 ij greatly enhanced" if RCA had had excellent terminals. (Tr. 56718- 

20; DX 2666, pp. 5, 15-16.) 

24 i 

25 

i 

-585- 



L' see also DX 2666, p. 5.) 

? During 1969 and 1970, RCA. looked at other ways to replace 

its Spectra Series: 

"One was to specialize in just one or a few industries 
and to market in no other than those few industries . 
Another alternative was to select an individual niche in 
the product line by rental size where we could market 
exclusively products in that area". (Rooney, Tr. 11820-21.) 

7 | At this point, during 1970, RCA had several alternative 

•:i 



a : 

3 

10 

il 
12 
-a 
14- 



16 
17 



ways it would proceed in the computer business : it could continue 
to market the Spectra 70 series until NTS , or some other more 
advanced product line was developed; it could specialize in a 
particular product area or it could market what became the RCA 
Series. The latter option — the RCA Series — was chosen for several 
reasons : 

(1) The new management of the RCA Computer Systems 

Division wanted to stop marketing the Spectra and to market 

its own line of products: 



"The then management of the Division wanted to 
have a product line that would be associated with their 
: j management era or period, as opposed to a product line 

IS ;j which was associated with an earlier management era". 

(McCollister, Tr. 9837-38.) 



19 



.: 



i 



il i 

•i 

12 ! 

t 









It also thought that a new product line would have a 
"psychological influence" on the "marketplace". (McCollister, 
Tr. 9816-17.) 

(2) RCA also believed that it could not continue to sell 
the Spectra series in the face of the price/performance 
improvements offered by I3M with its System/370. According 



-586- 



I 

2 

3: 

4 

5 

S 

7 i 

3 

9- 
10 
11 

12! 
13 j 

L4- j 
i 

15 ; 
i 

16 j 

17 I 
IS j 

19 j 
2Q ; 

21 j 

22 I 

24 : 
25 



to Rooney, if RCA had not been "selling against IBM" it could 
have continued to offer the Spectra series. (Tr. 12234-36.) 

(3) Because of its desire for a large market share RCA 

rejected the idea of focusing on particular product areas: 

" [A]t a meeting I attended, [A.D. Little] presented the 
concept that you had to have a broad product line because 
you could not possibly sell enough share of any particular 
product category to achieve this goal and that strategy 
was accepted as being valid" by "[t]he management of the 
Computer Systems Group as well as corporate management" . 
(Rooney, Tr. 11814.) 

(4) RCA believed that it could equal or better the 

price/performance of the IBM 370 systems and take away IBM 

users by introducing the RCA Series : 

"We were faced with a pending IBM announcement; we knew 
that the IBM announcement would offer their clients 
improved price/performance; we had just had the X series 
decommitted; and our objective was to grow to 10 percent 
share of the market. And we felt that we had to therefore 
maintain our original strategy of going after the IBM 
base. And after many discussions, it was concluded that.. 
by putting in the new memory capability we would be able 
to bring the cost of these systems down, so that we could 
offer a price competitive system — price/performance 
competitive system with IBM's 370. And since it was 
following the Spectra architecture, conceptually it would 
be the same strategy as IBM was employing, that is, 
utilizing the existing software for the next generation 
of equipment". (Rooney, Tr. 12242-43.) 

The RCA Series was announced on September 15, 1970, three 
months after the first announcement of IBM's System/370. (Wright, 
Tr. 13175-76.) The RCA Series consisted of four models "of small - 
to-medium-class computers — RCA 2, 3, 6 and 7". (PX 34 0, p. 17.) 

RCA described the RCA Series as "offering more power and 
memorv for the dollar than present third-generation svstems". (DX 675 



-587- 



I 



p. 16.) While the RCA Series had new memories, "under the covers, 

the RCA Series was essentially the Spectra 70". (McCollister, Tr. 

9819-20.) McCollister described the RCA Series: 

M [lit was a restyled prqduct line. There was a new set of 
covers,, the frames were the same, and it was essentially a 
cosmetic treatment of the existing Spectra 70 Series with 
new model numbers and new pricing. 

"There may have been some minor improvements. But 
'J fundamentally the product was not changed from the Spectra 
\ 70." <Tr. 9816-17, see Tr. 9819-20.) 



\ 
} 

1 

L S 

I I 
2 

& 

5 

3 

,7 ., 

'[paperwork and many presentations with respect to this rationale". 

:{ (McCollister, Tr. 9838.) Similarly, Rooney testified that in 1969- 
a J 

\ 1971 "we had a term called intercept strategy, which implied inter- 

*f* ! 
'Jj • 

! cepting the upward migration of the IBM client base with RCA equip- 

II S . 

jment". (Tr. 11811-12.) 

22 ; ! 

i The "elaborate strategy" failed.* The RCA Series was a 



RCA's computer division management devised "an elaborate 

strategy" to make the RCA Series succeed: 

"There was a very elaborate strategy at the time as to 
where these units of the RCA series would fall against the 
IBM either 360 or 370, either as it had been announced or 
was expected to be announced, and I think there was a 
fallacious expectation that in this elaborate strategy 
that the RCA series would fall at a certain point within 
the IBM product line spectrum and that IBM would be 
unwilling to disturb the equilibrium of that product 
spectrum and, therefore, negate the rationale of the RCA 
product concept." (McCollister, Tr. 9837-38.) 

Under this "elaborate strategy" the RCA Series would "intercept" the 

System/370. "Certainly there was a great deal of effort and much 



24 1 



* At the time of the decision Withington cautioned RCA about the 
25 | RCA Series. He testified: 

.1 

1 -588- 



"major product failure" and "a mistake" (McCollister, Tr. 9819-20; 
Wright, Tr. 13577-78; Withington, Tr. 56454-55.) The RCA Series 
failed in two respects: (1) instead of "intercepting" System/370, 
it "intercepted." RCA's own Spectra 70 series, and (2) it had sub- 
stantial technological problems. 

Interception of the Spectra 70. The witnesses used 
different words but all said the same thing: by introducing the RCA 
Series, RCA. "obsoleted" , "intercepted" and "blew . . . out of the 
water" its Spectra 70 series. (McCollister, Tr. 9838-39; Withington, 
Tr. 56720.) Withington explained what happened: 

"[A] great number of present Spectra 70 users renting . . . their 
systems, did indeed order the new RCA Series and indicate 
their intention to return their Spectra 70s, and that 
phenomenon indicated immediately that the financial impact 
of the new series would be more negative than planned"*. 
(Withington, Tr. 56711; see also PX 4836, p. 23.) 

So did McCollister:- 

" [A] customer had everything to gain by ordering an RCA 
series and returning the Spectra 70 . He got a brand new 



L 
2 

3 

* 

7 

a 

11 

12.1 

12 

14 

15 

is :! ' 
v J 

\\ "I remember indicating my concern to RCA management at 

XS \ that time about the likelihood that the new RCA Series would 

;j cause a general replacement of rented Spectra 70 machines in 

£9 : ;| the field, and that this would cause more negative financial 

;[ results than they were expecting. And I remember being 

2Q j uncertain about the degree to which IBM customers would be 

;! willing to convert quickly and in large numbers to the RCA 

2X \ Series". (Withington, Tr. 56710.) 



22 

22 



Withington "did not reach a firm conclusion at that time that 
failure was inevitable". He testified that the strategy might have 
been successful "if it had been carried out differently with different 
prices for the machines, with less effort to grew suddenly, and with 
2-4 ; j more effort on functional improvements in the product line". (Tr. 
56710-11.) 



25 

\ -589- 



machine. It cost him maybe 15 percent less or so and why 
not?" (McCollister, Tr. 9837-39.) 

The interception of the Spectra by the RCA Series seriously 

hurt. RCA in several respects. First, it reduced RCA's rental 

income because rents for the RCA Series were lower than for the 

Spectra. McCollister testified: 

" [T]he announcement of the so-called RCA Series of 
equipments as replacement to Spectra 70, which offered new 
equipment under different model numbers, which was technically 
essentially unchanged from the previous equipments , but which 
was offered at a lower price than previous equipments, meaning 
those equipments which were in that rental base . . . hastened 
the return of those equipments from the rental base to the 
manufacturer. 

"The effect was that it hastened the return of equipment 
that was then installed with customers on a rental basis 
because the customer could get a newly manufactured machine of 
equal ability, a new appearance and at a lesser price than he 
had been paying for the one which he already had installed" . 
(Tr. 11491-94.) 

Second, RCA was forced to build more RCA Series machines 

while it built up an inventory of returned RCA Spectra 70 's. 

Wright described the situation confronting RCA: 

"You therefore were confronted with the building of new 
products by the manufacturing organization and shipping those 
products to the installed customer base, and in many, many 
instances , because there was improved price/performance in the 
RCA Series over the Spectra series, you replace your own 
equipment and you got the Spectra series of systems equipment 
back. In many instances the cost of manufacturing had not 
been fully amortized, and that would have an effect upon both 
your cash requirements and also upon your P&L". (Tr. 13577-78.) * 



25 



McGollister testified that RCA found itself with: 

2.J "lo]rders for the RCA Series which required a manufacturing 

investment, in the product being placed out on rental for the 

3;| most part, which drew capital from the corporation to do this, 
and it resulted in the displacement of existing Spectra 70 

4. |f processors in many cases before they had been fully depreciated" 
(Tr. 9818.) 

5 

"This tended to build up an inventory of the equipment 

7 j| which was returned by the customers, the rental income from 
that equipment ceased and the company was faced with the 

g j requirement to invest money in new equipment to place in the 
customer's office to take the place of that which was sent 

g. I back or returned". (Tr. 11491-94.) 



IC 



He added that: 



In addition, the early returns of the Spectra 70s: 



*, "[h]ad serious adverse financial effect upon the Divison 

because it did not permit us to follow a plan or have a strategy 
•2 j which would maximize the return from the investment in Spectra 
j 70 equipments." (Tr. 9838-39.) 

13- ' 

14- 

15 

IS 

17 

13 

19 

20; 

21 l| 

I 

22 ;| 



55 i 

" , above) , by cessation of rentals from the returned machines and the 
• necessity to spend money to manufacture the new ones. 

25 



-591- 



' I 



The problems of returned Spectra 70 systems were such that 

RCA established a Returns Task Force, which made its presentation in 

early August 1971 and, considering both returns experienced to date 

and those which were forecasted, concluded that "approximately 70 

percent of the returns were being caused by RCA's replacement of 

Spectra series with RCA equipment" . Approximately 18 percent were 

losses to competition and about 12 percent due to economic problems . 

(Rooney, Tr. 12275-77; Wright, Tr. 13581-83; DX 873, p. 24.) These 

early returns particularly hurt RCA's profit and loss statements by 

forcing RCA to write off the undepreciated asset value of its "accrued 

equity contracts". (McCollister, Tr. 9820-21.) These contracts were 

arrangements in which the customer leased the equipment for five 

years, making equal monthly payments over that period, but RCA took 

70 percent of the revenue that it expected to achieve into its 

profit and loss statement in the first year of the contract. (Wright, 

Tr. 13589-96.) When the RCA series was announced, some equipment 

under accrued equity contracts was returned prematurely (according 

a ;| to McCollister) because the manager of the division "was anxious to 

make a showing with respect to the success of this new product line" 

and had "an inclination to allow customers to return Spectra 70 
t 
, j equipments prematurely for the sake of being able to cite an order 

^ i! for a machine in the new product line". (McCollister, Tr. 9820-21.) 

.- :« This meant that debits against current revenue had to be recognized 

J i 

, 4 \ when machines were returned before earning the revenue already 
»•» i 
I 

re l 






-592- 



^ i reported in prior years. 

2 ; " Technological Problems of the RCA Series. The RCA Series 

3 ; suffered front technological deficiencies that hampered its success: 
i • 

4-1' "Because you were: in a sense perpetuating technology 

J that was five years, old, you were making a new investment 
s ! in five-year old technology, and the pace of technology 
I in the industry, in it [sic] cost effectiveness charac- 
8 I teristics, is such that when you bring out a product line 

you cannot afford not to take advantage of improvements 
7 1 in cost performance and capabilities up to the time that 
you bring out that equipment. 

a !■ 

"That is certainly one aspect of the problem, that it 
9 | resulted .in new investment in old technology when better 
I technologies were available at that time". (McCollister, 
IQ | Tr. 9819-20.) 

I 
j^ i In addition to the general technological staleness of the 

! 

M J RCA Series, particular products were deficient in various ways. 

I 

£2 , Peripherals continued to be a problem. By 1970 RCA was "[t]wo to 

T<L ! three years" behind IBM in the development of peripherals.* (Rooney, 

• as L Tr. 12247-48.) More specifically, RCA was hindered by its failure 

t „ !j to have a disk drive competitive to the IBM 3330, which had been 

\ announced in June 1970 with System/370. In 1971 Rooney reported 

_ it that RCA still suffered from its: 

;| "[i]nability to provide a 3330 competitive device until some 
*£ ;| 19 months after IBM's delivery of its 3330 unit. I feel both 
, of these items are of major importance to the success of our 

20 j rca series marketing efforts and should be resolved. In par- 
I ticular, I am concerned we may suffer the same exposure we 

21 ! have faced with the 70/564 and 70/590 disc programs if we 
4 are not able to accelerate the present delivery schedule 

22;i for our 8580 unit." (DX 11101, p. .2.) 



2i j 

I * Rooney added that there had been a pattern in the RCA experience 

25 j up to 1970 of producing essentially carbon copies of IBM's peripheral 
! equipment two or three or more years late, either by developing that 
1 equipment themselves or by acquiring that equipment from other equip- 
,| ment manufacturers. It was one of his goals in 1970 to try to do some- 
! thing to overcome that disadvantage. (Rooney, Tr. 12249^50, 12252-53.) 

:: -593- 



► }•■ 



I i 



f i 



And in July 1971, RCA's computer division monthly report stated that 
"sales of the RCA 6 and 7 have been and will continue to be hampered 
by the large delivery differential between the RCA 8580 [disk drive] 

(March 1973) and the IBM 3330 (August 1971)". (DX 11099, p. 5.) By 
then RCA was arranging to purchase the 3330 on' an OEM basis from IBM. 

(See DX 937.) 

In mid-1970 RCA established a Peripheral Task Force to try 
to do something about its problem with peripherals. (Rooney, Tr. 
12249-53.) On July 16, 1970, L. E. Donegan, Jr., Division Vice 
President and General Manager of the Computer Division, wrote to 
W. w. Acker in Finance and Administration concerning the activities 
? I of the Peripheral Task Force: 



I 1 

8 



"Conceptually, the thing we must begin doing and you 
should attack it across the board in all peripheral areas, 
is get away from the pattern of producing carbon copies of 
IBM's peripherals, one generation late. Now that we have 
our hands on some very good IBM intelligence we should 
attempt to leap-frog and get ourselves to within 12 months 
of their delivered peripheral capability. 



. "... the tape units that we are presently developing 

' : in Marlboro are competitive in price performance and specifi- 
cations with Third Generation IBM tape stations and most 
® li likely will be non-competitive once IBM makes its new tape 

'( familv announcements. We must keep this from happening." 
9 1 (DX 862.) 



H 



RCA's problems with software also continued into the RCA 



- :; Series . VMOS 4 , an operating system to be used with the RCA 3 and 
;■! 

— j RCA 7, was announced in September 1970. (Rooney, Tr. 12335.) By 

.- :» 

a j December 1970 it appeared that there would be a 6-9 month slippage 

> 

-■* 1 in the delivery of VMOS 4. According to A. L. Fazio, RCA's Manager 
i 

£5 I of Virtual Memory Systems, such slippage would be a "product disaster" 



-594- 



1 
z 

5 

6 



causing RCA to lose about $3.5 million from delayed installations 
and approximately $2.1 million points (dollars in monthly rental) 
from current and future prospects. (DX 872 , pp. A, B, C.) 

The slippage occurred and was "significant". (Rooney, 
Tr. 12349-50.) Because of that slippage RCA lost $3.5 million in 
revenue and marketed about 40 systems — 20 RCA 3s and 20 RCA 7s — less 



j\ than it would have during 1571 through 1973.* 



(iii) Computer Systems Division's Problems — Early 1970s. 
1970 was "essentially" a breakeven year for the RCA Computer 
Systems Division. At the time RCA projected that 1971 would 
also be "breakeven", with 1972 showing a $25 million and 1973 
a $50 million "pre-tax operating profit". (McCollister, Tr. 9814-15.) 



9-! 
10 | 
U. \ 

i 

tt| 

■J3 || A five-year business plan drawn up by the Computer Systems Division 

in late 1970 provided for "a breakeven position in 1971". (PX 20 8, 

p. 1.) In its 1970 Annual Report RCA painted a similar picture for 

its shareholders : 



14 
IS 
16 
17 

IS IS 

20 j 

:i 

2 1 •! 
■ 

22 ; 

23 ! 

24 i 

2s ; 



"This investment has already resulted in a more rapid 
growth rate for RCA than for the domestic industry as a 
whole. In 1970, the value of RCA's net domestic shipments 
rose by more than 50 percent, while that of the industry 
fell by more than 20 percent. Among the factors responsi- 
ble for RCA's progress was a decision to continue increasing 
the computer marketing force during a period when many others 
in the industry were retrenching as a result of the weakness 
of the economy. 



* The purchase revenue on each RCA 3* sale lost would have been 
$1 million and on each RCA 7 sale lost would have been $2 million. 
(If rented, monthly revenue would have been 1/5 0th of the sales 
price.) (Rooney, Tr. 12352-53.) 



•' -595- 



I 

2 

3' 

4. 
•i 

5 

S: 

7 

3 

9 
L<3 
U 
12 
L3 
\A 
Is 

16 ;l 

17 j 



"Our computer revenue this year will be more than 
double that of five years ago, and we continue on target 
toward a profit crossover in computers in the early 1970' s". 
(PX 340, p. 3.) 

However, during the beginning and middle of 1971 it was 

« 

becoming apparent that those plans for the Computer Systems Division 
would not be met, and that the Division would be far less successful: 

(1) In late January 1971, Robert Sarnoff was informed by 
RCA's auditors, Arthur Young & Co., of a "major . . . change 
in operating results of the Computer Systems Division in the 
1971 business plan", and he asked for an analysis of the 
problem. Arthur Young responded by a letter dated February 
24, 1971. (DX 11108, p. 1.) 

(2) In April 1971 RCA revised its business plan for 
the Computer Systems Division. That plan reduced the pre- 
diction of revenue for the Division set forth in the December 
1970 plan from $323 million to $261 million. The revised plan 
predicted an. anticipated pretax loss in 1971 of $37 million. 
(DX 952, pp. 7-8, 12.) 



IS 

(3) On April 23, 1971, H. L. Letts, RCA's Senior 

19 ] 

:\ Financial Officer wrote to Sarnoff that the magnitude of 

20 : ! 

j the Division's problems raised serious long-term concern 

21 1 

j about the business and suggested reappraisal of the Division's 

22 ] 

objectives. He suggested that a task force be set up to study 

2a -i 

! the Division and its objectives. (DX 952, pp. 1, 2.) 

24 \ 

j (4) By June 1971 a task force comprising six persons 

25 -i 

1 from Arthur Young and the RCA Auditing Staff had reported 

: i -596- 



I 

2 

3 ! 

5 

e 

7 

3 

9- 
10 
LI 

12 | 

13 | 

14 ! 

15 ! 

j 

17 j 
IS 



on the problems of the Computer Systems Division. Morsey 
sent this report to Sarnoff and Conrad with a cover memo- 
randum stating that "the Computer Systems 1971 loss could 
deteriorate significantly from Business Plan levels" .. 
(DX 955,- p. A.} 

(5) During 1971: 

"[lit became apparent that there would be a loss 
in magnitude of $30 million or $35 million, a loss that 
eventually rose to the area of $50 million or $60 million, 
and this of course was an enormous difference from what 
had been anticipated at the beginning of that year . " 
(McCollister, Tr. 9814-15.) 

(6) By the middle of 1971 problems at RCA "put in ques- 
tion the anticipated revenue, and in turn opened the question 
in my mind as to » . , profitabilitv ... in the remainder nf 
1971". (Conrad, Tr. 14125.) 

Those and other participants in or observers of RCA's 
computer business pointed to many problems in RCA's Computer Systems 
Division that would cause its anticipated losses. A discussion of 
each follows: 

! 

i Declining Revenues. The higher-than-anticipated returns 



19 1 

: from Spectra equipment ($155 million as opposed to $90 million) 

• :! resulted in a reduction of expected net shipments (even though 

*•* 'I there was an increase in gross shipments) from $230 million to 

*- ij $186 million. RCA's 1971 Business Plan stated that: 

" Ij " [t]he returns implications of RCA's first introduction of 

ii ... the RCA series compatible with the Spectra series . . . 

2~ j were not fully reflected in the first plan. . . . There has 

; | also been greater migration than expected from the old to the 

~ w ;| new series. . . . The increased dependence on the RCA series 



■I -597- 

:l 
H 
ij 

Ii 



has a profound impact on revenue projections and attendant 
risk, since product will not be available until second half 
of 1971". (DX 952, pp. 7-8.) 

Expenses Too High. RCA's revised plan in April 1971 sug- 
gested possibilities of actions "to minimize corporate investment 
until development of larger revenue base is obtained", including 
the deferral of the Marlboro office building, "improved asset 
utilization and management", "further expense reductions" and a 
possible merger with or sale to another company. (DX 952, p. 21) 

McCollister attributed RCA's losses to the fact that 
"there was a substantial increase in expense, and that revenues 
were not increasing, and that revenues had been seriously over fore- 
cast. " An example was the construction of the Marlboro facility 
in 1971. "The relocation of the offices, the executive office and 
|j the construction of the office building under the circumstances was 
... a mistake. ... [It] was an expenditure which could have 
been deferred". (McCollister, Tr. 10964-65, see also 9814-16.)* 

RCA also "had a very serious problem relative to manufac- 
turing costs". According to Wright: 



.1 

24 1 * The expenditure for Marlboro amounted to about $25 million for 

•I capital facilities, not including relocation expenses. (McCollister, 

25 j Tr. 10966-67.) McCollister also testified that the 1971 opening of 

j RCA's sales office in the United Kingdom was "a mistake" because "it 
•\ was an investment which would, if ever, be financially rewarding only 
'I at some point in the future". (Tr. 10 96 5.) 

;l 
J 

i! -598- 



I 

2 



3 
9- 

ia 

II 

12 

13 
1* 
15 
16 
17 
IS 
19 
20 
21 



24 
25 



"There was no value engineering work going on after 
the product was developed to reduce its cost within the 
manufacturing organization. Those types of things . 

"The cost, as I recall it, when I first got involved 
in that, which would have been early in 1971 . . . was 
running* at that point in time about 42 percent of revenue." 
(Wright, Tr. 13559-60.)* 

Poor Organization and Unreliable Information. The 

Arthur Young report to Robert Sarnoff in February 1971 reported: 

"The basic failure to develop acceptable planning 
information in the division involved the lack of a reliable 
information base, principally relating to revenues, from 
which plans could be developed and current performance 
measured. This situation was aggravated by communications 
gaps which developed in a period of organizational change. 
Planning responsibilities and assignments were not clearly 
defined. As an example, the financial group was divided 
early in 1970; moves to upgrade the remaining group were 
less than successful. Preparing for the move to Marlboro 
was probably a further complication." (DX 11108, p. 1.) 

The Returns Task Force in 1971 reported that the Computer 

Systems Division suffered from the inadequate tracking of computer 

equipment : 

"1. No single, reputable data base for customer/equip- 
ment information. 

"2. No two data sources agree. 



* 



Wright recalled the comparable figure for IBM's manufacturing 
cost as a percent of revenue as "on the order of 14 to 15 percent" and 
"in certain other products, such as the CPU alone . . . substantially 
under that". RCA looked at other companies besides IBM and concluded 
ii that Sperry Rand's manufacturing cost "was running about 24 percent 
\ of revenue," and Burroughs "was about 21 percent of revenue." 
(Tr. 13560-61) When Wright got to RCA he "took several steps to 
reduce those manufacturing costs." (Wright, Tr. 13563) For example, 
RCA substituted "high quality plastic" covers for the "very heavy 
steel gauge covers" it was using at 15% of the cost. (Wright, 
Tr. 13564.) This and other cost cutting measures reduced manufac- 
turing costs by "8 percentage points as it would relate to revenue". 
(Wright, Tr. 13566.1 



-599- 



• :» 



L 
I j 

A 

■ ! 

.■i 

t 

: ! 



"3. Regular field inputs to data bases are clearly 
modulated by quota objectives bias. 

"4 . Recourse to the field for instant surveys 
leaves them short, on time, us long on dependence — in 
our survey we check out at about 85% accuracy. 

"Conclusions? Forecast based upon CS Data Bases and 
field surveys inherit a builtin error factor of +> 20%." 
(DX 873, p. 33.) 

Inadequate Financial Controls. A study of the Computer 

Systems Division in the summer of 1971 reported: 

"It has become apparent that CSD has not had adequate 
financial controls and analytical capability. Because 
of the complexity of the computer business in terms of 
revenue and cost forecasting, the interaction between 
generations of equipment, and the requirement for 
large, direct sales force, the control and analytical 
needs are greater in CSD than in most other businesses. 
These controls and analytical skills have clearly not 
been adequate in the division in the past. Moreover, 
despite some awareness of problems developing, Corpo- 
rate Finance did not provide the required support or 
leadership to the division in up-grading the controls 
and basic capability. If some of these problems had 
been made clearer earlier, the business might have 
been conducted in a different manner." (PX 349, pp. 6-7.) 

The lack of financial controls resulted in RCA's 

inventory being overvalued,* past due accounts receivables 



* The 1971 Study of the Computer Systems Division reported: 

"It has recently become apparent that a significant 
portion of Computer Systems inventory may be overvalued. 
Although it is not possible to identify at this time the 
extent of the problem, major writedowns will be required 
on video data terminals, Spectra 45 Mod 1, disc drives 
and other computer equipment. Available reserves may not 
be adequate." (PX 349, pp. 6-7.) 



-600- 



25 



* The report from Arthur Young observed: 



L i with significant amounts being prematurely written off,* and 

2 ; questionable orders being booked.** 

3 ! Product Deliveries . Because of product problems fore- 

4-1 casts of RCA Series shipments were not met. For example: 

5 [ "In the June 2 presentation, the 1971 business plan 

[ assumed shipment of sixty RCA 6 series in 1971. As of mid- 

S i July, Computer Systems indicated that the best estimate of 

RCA 6 shipments for 1971 was fifteen. A similar decline 
7 j has occurred in the case of the RCA 7. 

3 

9 

10 1 

"We believe that one of the most critical financial 

11 j problems facing Computer Systems is with its accounts 
receivable. Of approximately $27 million of billed 

12 | receivables, close to 50% ($13 million) are past due, and 
$3.3 million or 68% past due in Magnetic Products, Memory 

13 i Products and Graphic Systems. 

i 

14. Ij "This situation has been caused by many factors. A 

primary cause has been improper communication and incom- 

15 j plete data flow between the field and Headquarters , and 
within the Headquarters . 

16 11 

"Since many of the past due receivables are disputed 

U | items, a significant portion of the receivables are being 
written off or reversed, rather than cash being collected, 

ig I Credits to receivables have been averaging $3 million a 

jj month for the past twelve months ($37 million) . Receiv- 

12 j able reversals in May totalled $4.4 million, $2 million of 

;| which were netted directly against revenues. A limited 

2q '{ two-week test of Task Force collection results on receiv- 

•i ables past due over 90 days indicated that 40% were being 

27 ij reversed and only 60% collected in cash." (DX 955, p. 

~ ; 2; see also DX 11106, p. 1.) 

22 H 

23 ■ 

"A recently completed audit report indicates that 

24 ! one-third of the bookings represent questionable items. 

j ... This problem is compounded by the fact that a portion 
of sales commissions are paid at the time an order is 
booked and, therefore, salesmen may have been overpaid to 
the extent that the bookings are not firm. (PX 349, pp. 6-7.) 

-601- 



** The 1971 Study of the Computer Systems Division reported: 



"Despite assurances that time-sharing software 
problems had been solved last fall, software avail- 
ability continues to be a severe problem. ..." 
(PX 349, pp. 6-7.) 

< 

RCA Series Impact. The NTS series appeared to be likely 

to impact the RCA Series: 

"Based on expected introduction dates for the NTS 
series, it appears possible that a six-year life for 
the RCA series will not be achieved. A shorter system 
life would result in significantly greater write-downs 
in 1971 and future years. This impact could be antici- 
pated by increasing the obsolescence reserve or accel- 
erating depreciation but either of these actions would 
cause additional losses in the shorter term." (PX 349, 
p. 7.) 

Changes in Accounting Procedures . As noted above, premature 

; returns of products placed under "accrued equity" contracts forced 

RCA to take debits against current revenue. (Wright, Tr. 13589-91.) 

A draft release of the Accounting Principles Board "put in question 

the accounting practices being applied within RCA to the Accrued 

i Equity lease".* (Conrad, Tr. 14057-58.) The effect of a retroactive 

change in accounting practice would be large, involving a total effect 

i of a $53.6 million reduction in revenue for 1971 and a $104 million 

I 

j reduction in 1972 with a "substantial negative effect on the P&L 



4 



i * This was the same ruling which affected Telex and Memorex and 
r ! which would have required RCA to treat such transactions as leases 

1 rather than sales, ceasing its practice of taking 70 percent of the 
j, ! revenues to be received over five years as revenues received in the 

j initial year of the contract. (Wright, Tr. 13590.) 



-602- 



i i performance of Computer Systems in 19 71 . . . and even 

2. ; greater negative effect in 1972". (DX 956, pp. 5-6; see Conrad, 

3; Tr. 14058-65, 14069-70.) 

i 

t 

4-| The Economy. RCA's computer business was hurt by the 

i 

5 \ poor state of the economy in 1970 and 1971. "The economic 
i 

6 ! situation for the computer business in 1970 was quite bad. . . . 

7 | Shipments that year were down some 20 percent from the pre- 

i 
S| vious year." (Rooney, Tr. 12264.) The economic situation 

9 I increased the number of returns of computer equipment that 

i 

10, • RCA experienced. (Rooney, Tr. 12682.) The Returns Task 

t 
i 

Xl ! Force estimated that 12 percent of the returns of the Spectra 

\2 i in 1971 were due to the poor state of the economy. (DX 873, 

i 
13 i| p. 24.) 

! 

£4 ; Increased Competition. During the late 1960s and early 

i 
i 

£5 i 1970s increased competition hurt RCA's computer business. While 

i 

t 
IQ | RCA was putting out its old technology RCA Series , IBM was intro- 

yj 1 ducing a series based on new technology. As discussed 

above, prior to the announcement of that new IBM series 

RCA had attempted to predict the price/performance of IBM's 

anticipated new line in setting up its strategy. When the 

2t ;J 370 systems were announced in mid 1970, RCA found that its 

22 I predictions for the 370/155, 370/165 and 3330 disk drive were 

2- l "accurate". Its predictions for the 370/135 and 370/145, 

2* ! however, were "off target". RCA had anticipated that the 

price of the 370/145 would be 5 to 10 percent higher than the 



lS;i 

19 

20 



25 



-603- 



RCA 6 price; as announced, however, in September 1970 the 145 was 

priced approximately the same as the RCA 6. The 370/135 also came 

out with better price/performance than RCA had anticipated. The 

result of the inaccurate predictions were less sales for the Spectra 

6 and 2 because those systems were not as price/performance 

competitive as RCA believed they would be. (Rooney, Tr. 11910-11, 

11913-14, 11922-23, 12583-90, 12649-50; PX 349, pp. 6-7.) 

Moreover, Rooney testified that all of the significant 

technological innovations in 1970 were achieved by the IBM 

Corporation. (Tr. 12048-60.) Those included the 3330 disk drive, 

which "brought to users significantly improved price/performance, 

capability of storing and retrieving data on disks at much faster 

speeds than [previously]" (Rooney, Tr. 12049); semiconductor 

memory, which "reduce [d] the cost to the user in terms of the 

amount of money he would have to pay for memory . . . [and gave] 

the ability to have potentially much higher speed" (Rooney, Tr. 

m ! j 12050-51, see Tr. 11923) ; microprogramming, which allowed the 
t \ 

m ;| user to "improve the speed" with which he would process different 



3 



') applications and made it possible for comouters to more readily 
:{ perform the instruction sets of other comouters (Roonev, Tr. 

a i 

I 12056-57); and the 3211 high-speed printer. (Rooney, Tr. 

j 12058-59.) Rooney agreed each of the "significant innovations" 

! attributable to IBM gave IBM "a competitive advantage in 
3 1 



>e 



-604- 



Ij marketing commercial data processing systems." (Tr. 12060.) 

i 
i ■ 

Z\ RCA's competition was not limited to IBM. Compe- 

i 

3 S tition from other sources was also increasing. A "Computer 

i 

4- 1 Industry Survey"- prepared by RCA in February 1970 listed 102 

t 

5 companies (other than IBM) offering "computers", "periph- 

! 

S; erals and components", and "software and services". (DX 

i 

7 j 1110 7, pp. 4-6.) RCA was experiencing increased competition from 

3 | peripheral manufacturers. Wright, who was Chairman of the 

i 

9-j Peripheral Task Force in 1970, testified that the Task 

i 

IQ i Force was "surprised" and "shocked" by the number of users 

H | using, or intending to acquire, non-RCA equipment as part of 

12 \ RCA systems . This indicated to him that users "had learned that 

£3 i it was possible for them to achieve certain benefits by pro- 

| 
74.';' curing and mixing boxes from different manufacturers in 

* t 

£5 i the same system". (Wright, Tr. 13555-57; see also DX 852, 

i 

Tg } PP- 14 > I" 7 / 19-20.) By July 1971 RCA's Data Processing 

* i 

yj 1 Division monthly report listed among "significant problem 
areas": 



IS 
IS 
20 



"Independent peripheral manufacturers, i.e., Potter, 
Singer, have been waging extensive sales campaigns 
at selected customer sites. For example, Singer/ 
Frieden [sic] has proposed a plug to plug capability 

2«r l! for replacement of the 70/564 Discs at California 
~j Dept. of Justice." (DX 11099, p. 4; see also 

~ n \ Beard, Tr. 9021.) 

22 ; 

I 

22 j RCA also experienced increased competition from 

-. ! third-party leasing companies and the same July monthly 
-e ' report said that "discounts being offered on 360 systems 



-605- 



by third party leasing companies have [among other things] 
accounted for the slowdown on the demand for RCA Series 
systems".* (DX 11099, p. 5; see McCollister, Tr. 9290-92.) 

d. RCA's Decision To Sell Its Computer Business To 
S perry Rand. It was clear by the middle of 1971 that RCA's 
computer business had been hurt substantially by management 
errors, particularly by the introduction of the RCA Series. 
Yet it was not clear that RCA needed, or even wished, to 
sell its computer business. In 1971, according to Conrad, 
RCA's management had "a very strong commitment" to its com- 
puter business. Indeed, in July 1971, Conrad, speaking by 
videotape to a Computer Systems Division marketing manage- 
ment meeting, tried to dispel rumors that RCA would exit the 
business. He spoke "in substance" as follows: 

"As RCA's new President and Chief Operating Officer, 
let me assure you that the goals of your division, as 
enunciated so frequently in the past by Bob Sarnoff, 
are not changed. They remain faithful to his often 
expressed determination — to achieve for RCA a profit- 
able, long-range computer operation. 

"Every member of RCA's corporate management team 
stands behind that commitment. Contrary to rumors, RCA 



-\ 

1 

21 il . 

ij * McCollister testified that "the impetus for the concep- 

22 f tion and the development of the use of" the accrued equity 

ij contract, described above, "came about to some considerable 

23 I measure because of the presence of leasing companies in the 
j marketplace". (Tr. 9805.) Such contract began to be used 

2d ; extensively toward the end of 1969 to the early part of 

; 1970. (McCollister, Tr. 11038, see also Tr. 9804.) 3y 

2c 1970 between 50% to 60% of the new contracts written were of 

" : this tvpe. (McCollister, Tr. 9806-07.) 



-606- 



L 
Z 

z 

4* 

5 
a 
7 
3 

ia 

Ll 
IZ 
13 
14 



IS ; 



has no plans whatsoever to sell its computer operations. 
As we informed the New York Stock Exchange less than 
two weeks ago, these rumors are old hat. They've 
been circulated in the past. They were unfounded then. 
They are unfounded now* 

"Neither rumors nor setbacks will undermine our 
commitment ..to computers. We faced them before — as we did 
in color television. We believe computers can do for 
our company in the Seventies what color did in the Sixties. 

"On a personal note, I had an opportunity last 
week to express my personal commitment to RCA's com- 
puter business. I heard that rumors about RCA selling 
its computer business had caused the president of a 
mid-western railroad to revoke his order for an RCA 
computer. So I picked up the phone and called him 
directly. I told him exactly what I have just told 
you. And I've now been informed that the order has 
been reinstated." (PX 192, pp. 3-4; see Conrad, Tr. 
13939-40.) 

He added: "We are making a greater investment in the com- 
puter business than in any prior venture in our history. 
This is a measure of our confidence that RCA systems and 



._ [| products will effectively meet competitive challenges in 



the decade ahead." (PX 192, p. 2.) 

By September, however, RCA's view of its participa- 
tion in the computer industry had changed.* A group of 
executives consisting of Conrad, Sarnoff , Morsey and R. L. 



21 I * In Conrad's "judgment" the internal discussions in 
j RCA's management concerning staying or exiting from the 

22 I computer systems business "really began in August of 1971". 
j (Conrad, Tr. 13942.) On August 27, he and Sarnoff received 

22 ■:. reports concerning the status of RCA's computer business. 

| (See PX 201; PX 349.) 
24. i 



-607- 



r 
Y 

) 
I 

4 



Werner* (Conrad, Tr. 13942-44) met on September 16, 1971, for an 
hour and a half and decided to recommend to the RCA Board of 
Directors that RCA exit from the business. None of these four had 
ever had direct responsibility for the RCA computer division or had 
even worked in it. The Board of Directors adopted their recommenda- 
tion on the following day.** (Conrad, Tr. 13942-43, 13948, 14145.) 

The decision to sell the computer business came as a 
"surprise" to persons working in the Computer Systems Division. 
(Rooney, Tr. 12369; Wright, Tr. 13172, 13570-71.) RCA's management 
had not consulted with Wright, Donegan (Vice President and General 
Manager of the Division), Rooney or, to Rooney' s knowledge, anyone 
else in the Computer Systems Division. (Rooney, Tr. 12368-71; 
Wright, Tr. 13572.) 

The basis for the Board's decision to sell its 
business was that if anticipated losses of $137 to $187 



^ ' million in the Computer Systems Division over the period 

,7 ;: 

3 • 

j * Werner was RCA's general counsel. (PX 341, p. 38.) He retired 

•9 ; j as general counsel during March 1978 and continued as a director j 

;i until 1979, when he did not stand for reelection. (DX 13853, p. 41; i 

& j DX 13902, p. 48.) Conrad left RCA in 1976 (DX 13852, p. 37), and j 

i Morsey left in 1973. (DX 678, p. 40.) - j 

•j ** At the directors' meeting the recommendation was embodied in a j 
VL) memorandum (PX 208, p. 8), which Morsey read to the Board of i 



I Directors. (Conrad, Tr. 14072.) 



! 

.1 

It ! 



-608- 



L 
2 

3 

4- 

f 

7 

a 

3 

ia 

u 



1971-1976 materialized, that would cause a need for a 
greater investment in computers than RCA chose to make. The 
amount of the investment required over the 1971-1976 period 
was estimated to be $702 million. (PX 208 , pp. 4-5* ) The 
amount was disputed, however, even within RCA, and some 
j thought the figure was overstated by $100 to $200 million.* 

Competing with the computer division for invest- 
ment funds were the many other divisions in the RCA Corpo- 
ration. The needs of those other divisions for investment 
funds also were greater than had been expected; a "pre- 
liminary evaluation" showed that "new funds required" during 
1971-1976 "may exceed $1 billion" for the corporation as a 



i 
13 



» 



y, l * Julius Koppelman, the financial vice president of the 
Computer Systems Group, believed that the $70 2 million 
figure was in error by a very large amount. He, of course, 
had not been consulted before September 17. (Rooney, Tr. 

.„. 12370-71.) Wright testified that: 

la :t 

; "Julius Koppelman . . . subsequently had occasion to 
k * ij look at those financial numbers that had been put 

together by the corporate staff and submitted to the 
board, and Mr. Koppelmann told me that there had been a 

■j serious error in those calculations on the order of a 
£- | hundred million dollars, where it was overstated as 

,| far as the amount of cash that was required, and that 
22 j this had been caused primarily as a result of the 

| corporate financial people not understanding certain of 
21 j - the contractual terms and conditions that we had with 

A customers on certain of the equipment that we had 

22] installed." (Tr. 13572-73.) 



23 j Similarly, Rooney testified that he was told by Donegan 

j "that Mr. Koppelmann . . . told him that there was an error 

2A ) of some $200 million in those numbers ... in overesti- 

! mating the capital requirements of the business." (Tr. 

25 j 12371-74.) 



-609- 



whole. (PX 208, p. 5.) 

It was against that background that RCA made its 
decision. It considered whether to proceed with the magnitude 
of investments contemplated both in computers and other 
areas, and believed that "if earnings growth can be maintained 
!i at an annual rate of 10%-15%, the Company can raise needed 
funds". However, if RCA earnings were only to grow "at a 
rate similar to GNP (7%)" or if a recession were to occur 
"the resulting reduction in RCA's overall profit position 
could bring considerable pressure on obtaining the $1 billion 
outside financing required". There could be "severe financing 
problems". Major losses in computers would add to the 
difficulties. (PX 208, p. 6.) 

Conrad testified that he believed that RCA could 

have raised the necessary capital to finance the projects 

that the Computer Systems Division had in mind at the time, 

■[ could have reached its goal of achieving 10% of some 

'•I 

I defined market, and would at some point in time have been 

l| profitable. (Conrad, Tr. 14047-48, see also Tr. 13944-47.) 

j Similarly, Rooney — who as a member of the Computer 

T ! 

•| Systems Division had not been consulted prior to the decision — 

1 'j ■ 

■j testified that he believed that RCA could have been successful 

2 ••; 

| in displacing IBM products in the 1970s had it been allowed 

3 i 

to continue in the business. (Tr. 12094.) 
A i 

j Withington, who had advised RCA's Computer Systems 



-610- 



I 
: 

3 

7 
I 

3 I 



L| Division a year earlier, believed that RCA could have been 

2 ; successful in the computer business had it chosen a differ- 

31 1 ent, less adventurous, strategy and remained profitable while 

4-|* growing more slowly. (Withington, Tr. 56711, 56720-21.) 

i ■■ 

5* i ; RCA considered what actions could be taken to 

>< 

I' 

S. J! mitigate the risk that it would need a large amount of 

7 j| financing if it stayed on its present strategy in computers. 

3 I One alternative mentioned was to 

g; |; "reduce investments in other new businesses such as 

i the Supermarket project, the French Color Tube venture, 
£Q •' SelectaVision, etc. , thereby substantially reducing 

i. outside financing requirements. Even in this case, if 
j^ (■ profits were to grow at 7%, or if a recession were to 

I occur, it is likely that the major financing which would 
n \ still be required for CSD could only be obtained at 

| higher interest rates and more restrictive terms . In 
£3 ; addition, RCA would in this case be relying even more 
directly on the ultimate success of CSD by passing up 
T( l '} other, perhaps more attractive opportunities." (PX 208, 

p. 6.) 



is ;i 

17 



Another alternative would be to keep the computer 
business but "to substantially reduce the size and scope of 
the computer operation" by limiting it to certain narrow 

-^ it market areas. However, it was believed that this would 

tc 1 

iS ;} reduce revenue as well as expenses and "while cash requirements 

would be reduced substantially, it is questionable whether 

the business would ever attain economic viability". ( Id. , 

p. 6.*) As a result, "the additional' investment required in 



& ■! 



H .1 



L.Z. ■( 



4A \ 

! 
i 

\ 

24 1 



* Other alternatives mentioned were to cut dividends and sell 
surplus real estate and "marginal business". (PX 203, p. 6.) 



-611- 



\ 
} 

1 

I 



CSD no longer appears to be a prudent financial risk" . 
( Id. , p. 7.) 

A major concern was that if there were a downturn 
in the economy in the mid-1970s r "the many healthy and vital 
parts of the rest of RCA could be hindered" because of the 
"high level of outside financing" required for computers 
"as well as the other parts of RCA" . "While the computer 
industry is an attractive and growing business — although at 
a slightly lower level than originally anticipated — the 
profits to be gained by RCA from this business, in relation 
to the total investment required, do not appear to be con- 
sistent with sound financial planning" . ( Id. ) 

In addition, the memorandum said that given RCA's 



\ position in the computer business "[t]he manpower and finan- 
cial resources of IBM, including the size and strength of 
the marketing, research and development organizations, are 
such that achieving market share growth as well as accept- 
able profitability, is extremely difficult". (PX 208, 
p. 7.) Thus: 



4- 



:t 

si 



» 
i 



"In summary, the computer business currently 
accounts for about 6 percent of RCA's total revenues. 
While it could represent a growing segment of the 
LI | Company's operations, it is unlikely to ever exceed 

■j perhaps 10-15 percent of total RCA volume. Continued 
:2 ;l commitment to computers, however, could lead to severe 

j financing problems for the Company and may contribute 
23 | to restricted growth in other operations. On balance, 

! it is believed that the risk does not justify the 
l*. j potential reward. Therefore, withdrawal from the 

j mainframe computer and peripheral equipment business is 
?« I recommended." (PX 208, p. 8.) 



-612- 



2; 

i 
^ i 

1 
i 

_ ! 
3 i 

! 

5 j 

3 | 
9 I 

1C j 

11 ! 

i 

12. 
13 

15 
16 
17 
IS 



Similarly, Conrad testified that RCA left the 

computer business because: 

"[Wje believed that we could apply the resources 
employed in the computer business better in other 
opportunities within RCA, which would lead to 
profit at an earlier time with greater assurances." 
(Conrad, Tr. 13933-34; see also. PX 217, pp. 2-5; 
Conrad, Tr. 13989-94.)* 

All the participants and observers of RCA's deci- 
sion to sell its computer division agreed that RCA's deci- 
sion was voluntary, and "nothing IBM did or any other 
company forced us or caused us to exit the business per se" . 
(Conrad, Tr. 14046; see also PX 217, p. 5; Rooney, Tr. 12387; 
Wright, Tr. 13630-31.) 

After September 17, 1971, RCA negotiated with 
Sperry Rand and with Mohawk Data Sciences concerning the sale 
of its computer business, as well as having meetings with 
several other companies , including Burroughs , Xerox and 
Memorex. It sold the division to Sperry Rand.** (Conrad, 
Tr. 13953-54, 13968-70; PX 402.) In its presentations to 



•t 
:l 

Q 



ij * Among the projects in which RCA believed it could 
2G j invest its money with more opportunity for profit than the 
i computer business, the principal ones were Alaska Communi- 

21 ; | cations, the Global Communications Company, SelectaVision, 
•land VideoDisc. (Conrad, Tr. 14046-48.) By 1978 RCA had sold 

22 j Alaska Communications. (DX 13854, p. 15..) 



22 ;| ** During the period subsequent to 196 6 RCA also withdrew from 
many businesses in addition to the business of its Computer 

4* Systems Division. Those businesses include: the manufac- 
turing and marketing of radios, phonographs and stereo equip- 

25 ment; the manufacturing of radio tubes; a color picture tube 



-613- 



prospective purchasers, RCA estimated that the "After Tax 
Cash Contribution" of its lease base for the period 1972 
through 1974 would be $193 million assuming no residual 
value. (PX 405A, p. 8.) It sold its computer division 
to Sperry for approximately $137 million. (Conrad, Tr. 14130.) 

RCA reported that it lost approximately $241 mil- 
lion before taxes on its computer systems operations for the 
years 1958-1971*. (PX 410.) 

f. j In September 1971 RCA set up a reserve of $490 mil- 

i 
j ; lion pretax, $250 million after tax, to cover prospective 

losses in connection with the sale of its computer division. 

The losses that were anticipated related to disposition of 

• i assets "such as inventory, receivables, plant", "discharge 

of claims and obligations for commitments to employees for 






>«• ! 



:. joint venture with Thorn Electrical Industries; Meyer Bros. 

i Parking Services; United Exposition Services (a subsidiary 

' ; of Hertz that engaged in services related to exhibitions) ; 

:| Cushman & Wakefield; the design, manufacture and sale of 

* ',i microwave communications transmitters, receivers and multiplex 

:{ equipment in the United States; electron microscopes; 

- :i Service America, which offered to service televisions of all 

:j manufacturers; RCA Alascom; and Random House. By 1980 Banquet 

2 ,j Foods was also up for sale. (Conrad, Tr. 14022-27; DX 13854, 

t j p. 15; DX 13860, p. 8; DX 13902, pp. 2, 36-37.) 

\ * RCA's losses in the late sixties had "to do with the 

— .! investment that we felt we had to make and the engineering 
;j and programming for future profitability in order to grow 

2 | in *he business", as well as to RCA's accounting for leases. 
" [Pjrofitability was governed primarily by the rate at which 
RCA determined that it would like to grow", about 20% a 
year, "somewhat faster than the general growth of the market". 



5 ! (Beard, Tr. 8535-38. 



-614- 



L | 

2 | 

i 

4-i 



3 

9 

II 
12 

13 

L4 

16 

i 
IS 

19 :l 



severance and release", and other purposes. In December 
1973 a review- indicated that the disposition was going better 
than expected and the reserve was reduced by $78 million, 
leaving a pretax reserve of $412 million. (Smith, Tr. 14247-48.) 

e. After the Sale to Sperry Rand. The story of 
RCA's participation in the computer industry after the 1971 
sale to Sperry Rand has two parts: RCA's activities and 
Sperry Rand's success with the computer division it purchased 
from RCA. 

(i) RCA's Activities. As it had planned, RCA invested 
heavily in other businesses in the 1970s. Conrad testified 
that from 1972 through the end of 1976, RCA invested approxi- 
mately $130 million in satellite communications (Tr. 14009-10) , 
more than $250 million in the same period in Global Communica- 
tions (Tr. 14011-12) , approximately $200 million annually in 
the purchase of automobiles for Hertz (Tr. 14102-06) , and $150 
million in Alaska Communications. (Tr. 14008-09.) 

j RCA also continued in or entered computer-related 

i 

i businesses. Conrad testified that: 



j "In our Solid State Division, we manufacture, design, 
^ j engineer and manufacture integrated circuit devices 
called microprocessor chips, which can be and are 
used in data processing applications, as well as com- 
munications applications. 



21 •! 

•4 



"We also continue to offer and perform service 
22 j on a variety of data processing or reservation system 

j terminals, which are owned by others. 

t 

J "We continue to from time to time design, develop 



-615- 



and manufacture special processors which are sold to 
the government in conjunction with tracking devices, 
such as radar." (Tr. 14048-49, see also Tr. 14157-58.) 

(ii) Sperry Rand's Success with RCA's Computer Systems 

Division. Sperry believed that its acquisition of RCA's 

Computer Systems Division was "a sound business" and a 

"wise" decision. (McDonald, Tr. 3873-74; DX 63, p. 1; DX. 71, 

p. 9.) In its 1973 Annual Report Sperry reported: 

"More than 90% of these RCA customers remained with 
us, and more than $130 million in new equipment was 
shipped to these users during calendar year 1972. 
We are continuing to build ' bridges ' between the RCA 
systems and Sperry Univac's line, and we are confi- 
dent that many of these customers will eventually 
convert to Sperry Univac's systems." (DX 63, p. 1.) 

In December 1974, 77% of the original RCA customers acquired 
by Univac were still using their RCA equipment and 5% of 
L ;j the original customers had moved to Univac systems , and the 
RCA equipment had yielded a "revenue stream (sales, rentals 
and maintenance) for 3 years of approximately $370 million". 
Univac believed' that "these benefits will certainly not end 
at this point". (DX 68, pp. 11-12.) By May 1975, approxi- 
mately 76% of the RCA equipment acquired by Univac was still 
on rent. (McDonald, Tr. 4045-46.) 

f. Conclusion. Like General Electric, RCA was a 



k ' 



7 
3 

7 



3 



T 



} large company with a small computer business. In the 
2 -| 

last full year before its sale to Sperry Rand, RCA's U.S 



1 



EDP revenue was $226 million. (DX 8224, p. 2.) RCA's 
venture into computers was a failure; but it need not have 



-616- 



L j been. As we have seen, despite RCA's great technological 
2. | capability in the 1950s, RCA only placed nine computers in 
that decade. RCA's inactivity in the 1950s and early 1960s 
4*jf cost them, dearly. But RCA still could have succeeded. The 

5 [ : Spectra, patterned after IBM's 360 was a mixed success. 

I 
a i Reliability problems and inadequate peripherals limited the 

7 ;' acceptance of the systems. But even then had RCA understood 

g || the need to push ahead with technological development, to 

g. I commit its ample resources to new, more advanced follow-on 

iq ! systems, it could have succeeded. 

i 

££ ! Instead, it introduced the RCA Series — yesterday's 

£2 j technology at lower prices — and it chose that vehicle to 

! 

£3 I spearhead its drive to "gain market share" and "become number 
T< # : 2 in the industry". But the RCA Series could not compete 
£5 1! with the more advanced products of IBM and others and was a 
"major product failure", blowing the Spectra series "out of 
the water" . 

At the same time, the management of RCA changed 
hands and the company sought to transform itself into a 
conglomerate. The result of its conglomeration was that 
all the various corporate mouths needed feeding at once 



16 il 

i 

is; 
is! 



20] 



?7 



•i and as the company entered the recession of the early 1970s, 
it found itself stretched too thin to pay adequate attention 



23 



4A I 

-I 

t 
i 



or commit sufficient resources to save the computer business 
from the RCA Series debacle. 



-617- 



r j 
i 



I 



I- 
: 

m 

3 

A 

.1 



° 



? :l 



2 \ 



W- i 

i 



IS i 



In sum f the story of RCA, like the story of 
General Electric, is the story of missed opportunity, bad 
management and product failures. 



•\ ! 



-618- 



2 

5 | 



7 
S 



43. Honeywell. The history of Honeywell during the period 
1964-1970 was built on the success of the Honeywell 200 — a product 
which gave birth to a compatible family of computer systems and in 
r turn sparked expansion of Honeywell's peripheral line and service 
capabilities. Despite some difficulties along the way, Honeywell 
ended the sixties with a large and successful array of electronic 
data processing products and services with rising revenues and profits 
derived from them. 

a. The 200 Series. In December 1963 Honeywell had announced 
n !J its 200 computer system, along with an "automatic program conversion 
n | package, called 'Liberator'". (DX 198, pp. 25-26.) Richard Bloch, 
: who led the Honeywell team that designed the 200 (Tr. 7886), testified 

: to the strategy behind it:* 

i 
14; 

j ; * Richard Bloch, who was Vice President for Product Planning at 
^3 j Honeywell for most of the period he was there, 1955 through 1967, 
f testified about Honeywell. His duties as Vice President were to 

16 ;l develop product, pricing and marketing strategies for the products 
•of the EDP Division. (Tr. 7575-76.) 

17 I 
: James H. Binger and Clarence W. Spangle were the other witnesses 

IS 'I who testified about Honeywell. James H. Binger was Chairman of the 
•| Executive Committee of Honeywell, Inc., when called to testify. (Tr. 

19 j 4489 . ) An employee of Honeywell since 1943, he became chief executive j 
'} officer of Honeywell in 196 4 and held that post for the following j 

20 1 decade; starting in 196 5 he was also Chairman of the Board of Directors. j 
| (Tr. 4489-90.) | 

21 it i 

\ i 

4 Clarence W. Spangle was President of Honeywell Information Systems j 
22. : | and Executive Vice President of Honeywell', Inc. in 1975. (Tr. 48 82.) j 

| Spangle, too, had been with Honeywell since the 1940s. ( Id. ). From I 
23 1965 through 1969, Mr. Spangle was Vice President and General Manager j 

i of Honeywell's Electronic Data Processing Division, responsible for j 
24. [marketing, manufacturing and development of data processing systems. ! 

.} (Tr. 4887-89.) 



-619- 



L 
2 

3 

t 

i 

Si 

i 

i 

7 j 

i 

3 i 

i 

i 
I 

ia : 

i 

n ! 

12 

13 
14* 
15 
16 



"[T]he 200 was conceived to represent a next step for 1400 
Series users in the IBM line, and we really designed that 
machine from the outset to be attractive to that user community. 

"One of the attractive features had to be a means of 
getting the user to bri,ng his programs on to the new machine, 
and that required conversion facilities and conversion 
offerings, which we intended and indeed did develop." (Tr. 
7578.) 



Honeywell felt this strategy would give it an "accelerated move into 
the [general purpose business data processing] field, which we needed." 
(R. Bloch, Tr. 7585-86.) In 1963 Honeywell's EDP revenues were only 
4% of its total revenues, and it had yet to make a profit in that 
area. (DX 132, p. 11; DX 198, pp. 4-5; DX 8224, p. 387; DX. 14484, 
p. Rl; DX 8631, pp. 31, 37.) 

The 200 was designed to make conversion from the IBM 1400 
series as easy as possible. (Binger, Tr. 4823; R. Bloch, Tr. 7886.) 
An effort was made to replicate closely the file structure, media and 
formatting of the 1400 (R. Bloch, Tr. 7605-06; Spangle, Tr. 5025) and 
the LIBERATOR conversion aid was developed. (R. Bloch, Tr. 7606.) 
U ;j The LIBERATOR enabled 1401 programs to be converted to 200 series 
13 i programs by means of assembly language and object code translators, 
ig 'J ( Id. ; Goetz , Tr. 17652-53.) Because the conversion required only 

20 ] a very small amount of manual intervention, it resulted in a high 

21 jj degree of efficiency. (Spangle, Tr. 5021-22; Goetz, Tr. 17652; see 
22 ; ialso R. Bloch, Tr. 7888-89.) 

22 ! ! The LIBERATOR successfully accomplished the conversion for 

■I ! 

2* : |which it was designed. (Binger, Tr. 4823; R. Bloch, Tr. 7888-89; see i 

■i ! 

?c '-also Goetz, Tr. 18780-81.) Thus, the 200 offered users both an easv | 

I I 

-620- ! 



L 
Z 

3 

7: 

3 
9 

ia 

u 

12. 



conversion method* and price/performance superior to its competitors, 
including the 1401. (McCollister, Tr. 11237, 11365-66; Evans, Tr. 
101187-88; PX 6204, p. 4; DX 167.) Both of these characteristics led 

■ the Honeywell 200 to enormous success. (R. Bloch, Tr. 7602-03, 7888; 
McCollister, Tr. 11235-36; Withington, Tr. 55863-67; J. Jones, Tr. 
78989-95.) Withington wrote in October 1964 that Honeywell had 
obtained many hundreds of orders 'for the 200, and that no computer 
manufacturer was gaining ground as fast as Honeywell. (PX 4829, pp. 
20-21.) IBM reports called the 200 an "outstanding success", 
allowing Honeywell to expand its marketing and other personnel and 
to turn a profit for its ZDP Division. (PX 3481, p. 69.) 
McCollister testified that Honeywell expanded its sales force during 
the early sixties so that by 1965 Honeywell had 50 to 75 percent 
. , ! l more people than did RCA, although the two companies had started the 

I sixties equal. (Tr. 10962-63.) Sales were made to all kinds of 
customers; Gordon Brown testified that as a Honeywell marketing 

; representative he sold 200s to an insurance company, an aircraft 

1 company and a service bureau; Honeywell reported sales to the Internal 



12 



13 
16 
17 
IS 



it 

; j Revenue Service and to U.S. Air Force Major Air Commands. (G. Brown, 

Tr. 50985; DX 13349, d. 27.) 

2Q .1 

i Within IBM, the Honeywell 200 announcement provoked heated 

21 J . 

i discussion on how soon IBM was going to .come up with a better per- 

22 ;; 

| forming product with which to respond. Immediately following the 

23 1 



25 .1 * The 200 was also compatible with "most widely used 
| small computers". (DX 167; DX 198, p. 26.) 

-621- 



1- i announcement, T. V. Learson, IBM's Senior Vice President, wrote to T. 
2 ; J. Watson, Jr. , its Chairman, and A. L. Williams, its President, that 
3" ! the Honeywell announcement was "even more difficult than we antici- 
4*1 pated". (PX 1079.) Shortly thereafter, M.. T. Hague, IBM Director of 

i 

5 I : Market Programs, wrote to Dr. J, W. Gibson, IBM Vice-President and 
5-i Group Executive, that the 200 "represents the most severe threat to 

i 

7 j IBM in our history". (PX 3912.) Evans testified that the marketing 

i 

! 

8 'force and others dealing with the 200 regarded it as a real challenge 
9; to IBM (Tr. 101186), and both he and Knaplund testified that the 

IQ : marketing organization put a lot of pressure on the development 

I 

11 j organization to announce the 360/30 as soon as possible in reaction to 

i 

12 I the 200. (Knaplund, Tr. 90475; Evans, Tr. 101190.) On January 28, 

13 j W. C. Hume, President of the Data Processing Division, wrote to Dr. 

i 

14. ; Gibson: 

t 

15 i "We must have an answer to this system immediately .... 

| The best solution to this problem ... is a 101-H [360/30] 

15 ! machine with a competitive price to the H-200 and a 

ii performance equal to or greater than the H-200, ready for 

17 I announcement by mid-February." (PX 1090; see pp. 353-57 above.) " 

13 ; Of course, by April 1964 IBM was able to respond to the 

iq | H-200 with the 360/30 — which was two or three times more powerful 

I 
2Q j than IBM's 1401 at less than one and a half times the price. (Hughes, 

Tr. 33924; JX 38, p. 33; DX 573, p. 6; see pp. 230-31 above.) Despite 



21 

22 •; the announcement, Honeywell continued its successful course. In 

22 ; ! December 1964 T. V. Learson wrote to T. J. Watson, Jr. that "the 
25 j 

i 

1! 

-622- 



L ; Honeywell 200 story" had led to 300 losses to date for tape-oriented 

2 ! systems, with 1,000 such situations in the doubtful category, 40% of 

3 !. which he estimated as losses. (PX 1288, p. 2.) 

4» V Honeywell spent the remainder of the decade enlarging on 

it 

5" if and solidifying the 200 's success. In a 1969 speech, Binger outlined 
his strategy: 



3 



T 
3 

10 
11 

12 
13 
L4- 



16 
17 
IS 



"In the beginning we made a conscious decision and adopted a 
strategy to compete in a broad segment of the computer 
marketplace, and to make significant penetration through a 
wide array of products and services . Our highly successful 
Series 200 computer line is the prime example of this 
strategy." (DX 132, p. 12.) 



In June 1964, after IBM's announcement of System/360, 
Honeywell announced the 2200 and followed it in February 1965 with 
three other compatible new members of the Series 200, giving it a 
"family of computer systems": the 120, the 1200, the 2200 and the 
4200. Honeywell stated that "[t]he family concept of these new 
systems gives our customers the assurance that they can meet 
j problems of growth by expanding through an extended range of central 
| processors , continuing to use the peripheral equipment already in 
i their EDP system."* (DX 13849, p. 27.) The 200 series 
** | was also, through hardware design and programming adaptations, 
22 ,j "accessible to [Honeywell's] 400 and 800 [users] who can shift to 

~ 1 the higher levels of the newer series with a minimum of adjustment, 

■•i 



I and with the protection of a substantial part of their prior 



23 



Z- \ * The compatibility of the 200 family meant each model was also 

J compatible with the IBM 1401. (Withington, Tr. 56375-76.) 



7« 



-623- 



r 

1 \ 

i 

i 

3 i 

L 
1 

2 » 
4- 



6 , 

7 i 
S 
9 

:c 



programming and file investment." (DX 199, p. 32.) 

Spangle testified that the 200 line was priced so that the 
three-year lease prices would be "roughly equal to those of IBM for 
equivalent price/performance on a system basis", with the one-year 
r if price "slightly above that of IBM" and the five-year price "5 to 10 
percent below the one-year price of IBM" . The consideration of what 
the price should be was based on the price for the system, although 
the individual elements of the system were priced individually. 
(Spangle; Tr. 5056.) 

According to Richard Bloch, Honeywell gauged its pricing 
"against the nearest competitive IBM line or the IBM equipment which 
we were hoping to supersede, to a lesser extent some of the other 
competition" because "if we were to increase our penetration of the 
market we would obviously have to take away some of the captive 
business that was presently in IBM's hands". (Tr. 7596-97.) Honey- 
well priced its product so that it demonstrated performance advantage 
over IBM "that might be measurable in tens of percent" with a price 
equal to or less than the IBM price. Where, on the other hand, 

i 

'j Honeywell felt that it did not have any substantial performance 
:| advantage, it considered that it would certainly have to have a 
i significant price advantage* This meant 90% or less than the IBM 



"" '•• i 

.j price. However, there was no "automatic rule of thumb". (R. Bloch, s 

12 'I j 

I Tr. 7599-601.) I 

S | 
As IBM improved the capabilities of its 360 line (see pp. ' : 

U ' J 

f above) , so did Honeywell with further improvement of the capabilities 02 

15 "j ! 

] i 

1 -624- I 

: ! I 

i • ! 



the 200 family through peripheral and software announcements. A number 

of new products , both hardware and software , were announced for the 

Series 200 line at the end of 1966, covering mass storage, data 

communications and expanded multiprogramming, including four magnetic 

disk devices for "random access information storage and retrieval"* 

and a number of terminal devices. (DX 199, p. 31.) 

The tempo of announcements accelerated as the decade went 

on. In its 1967 Annual Report, Honeywell stated: 

"Product lines of the Electronic Data Processing and 
Computer Control Divisions [**] are being broadened con- 
tinuously to assure competitiveness. In 1967, for example, 
more than a hundred hardware and software products and 
product modifications were added to the Series 200 EDP 
line and the control computer line." (DX 200, p. 31.) 

Binger and Bloch testified that Honeywell's competitors in 

the 1960s were IBM, Sperry Rand, RCA, Burroughs, GE, CDC, NCR and 

DEC. (Binger, Tr. 4527, 4593-94; R. Bloch, Tr. 7592-94; see also 

Spangle, Tr. 4933-34.) SDS was a competitor "to a limited extent". 

(Binger, Tr. 4515.) Honeywell also faced efforts by Fujitsu, Philips 

and Nixdorf to sell their computer equipment in the United States . 

* II (Binger, Tr. 4516-17.)/ 



J ! 

I 



;j * These were acquired OEM from CDC. (See below, p. 628.) 
- is 
j ** See below, p. 633. 

/ Other foreign competition was encountered, too; for example, Honeywe 1 ] 
bid against Siemens, among others, in 1968, for an accounting and j 
, payroll system for the U.S. Army. (DX 7556.) j 



-625- 



i : 
\ 

l 

I 

l ! 

1 

t 

S 
7 
S 



si 

•l 



Honeywell's 200 series was sufficiently popular to be 
marketed by leasing companies. Thus, Leasco dealt in Honeywell 
equipment before 1967 (Spain, Tr. 88749) and Finalco was leasing 
Honeywell 200s and 1200s, a fact which made Patton, a Honeywell 
Regional Sales Manager, "a little nervous over what could happen if" 
those systems come off lease". (Spangle, Tr. 5191-92; DX 161A-) 
Transamerica also leased Honeywell equipment in the late 1960s. 
(Spangle, Tr. 5190.) 

b* Problems and Solutions. 

(i) Other Systems . Honeywell was not without problems over 
this period, though. One of these was the 8200 computer system, 
which was planned to be the most powerful computer system in the 
200 series. (DX 13849* p. 27.) The Honeywell 8200 was announced 
in 1965. ( Id. ) It was intended to bring together the Honeywell 
200 line and the Honeywell 800 line, the latter of which was in- 
stalled at that time at about a hundred different sites. (Spangle, 
Tr. 4997.) 

"At the time of the announcement the development of the 
machine had not begun. And as the development was undertaken, it 



■) turned out to be much more difficult to do those things than had 

a .! 

■j been anticipated," and Honeywell spent large amounts of money, more 
'1 } . 

j than it had planned, to develop the equipment and to develop the 

j software to supply with the equipment. ( Id. ) 

j Honeywell was not able to achieve the objective of having 

?.4 'i 

-"* I 

.{ that system be an upgrade path for the 200 line so, according to 
Spangle, "its market became limited really to those 800 customers 



-626- 



L ! 

i 

2 ' 

i 

i 

i 

* I 

I 

5 I 

7 

3 

9 
10 
11 
12 
13 
14 
15 
16 
17 
IS 



who wanted to continue largely in the batch processing mode and 
wanted higher throughput". (Spangle, Tr. 4997-98.) 

As a result of all this, Honeywell was able to ship only 
about 40 of these machines which Spangle testified was not enough to 
make the whole investment and development worthwhile. The particular 
problem that caused the 8200 to fall short of its objective was the 
need for two operating systems in one computer system — Honeywell 
could not get it to work. (Spangle, Tr* 4998.) 

Honeywell tried to aid its customers with 800 systems 
installed in another way — by the provision of a larger system which 
was compatible only with the 800. Thus, it announced the. 1800 in 

1962. (See p. 189 above.) However, according to Withington, 
the 1800, designed to accommodate the growth needs of the 800 
users, "sold in only very small amounts". He attributed this to the 
fact that the IBM 360 and GE 600 series, available at the same time, 
"were regarded as superior to the Honeywell 1800 by users and 
Honeywell users who outgrew their Honeywell 800 apparently more 
frequently left Honeywell for a competitor than accepted the 1800 



T q If instead" . The 360 and GE 600 systems were felt to be superior to 
3g ; | the Honeywell 1800 "because they offered early versions of operating 
2i :! systems whose primary initial virtue was to permit multiple pro- 
22 I' gramming . . . plus automatic control of peripheral equipment in 
~~ 'I ways which would simplify the users' programming requirements". 

: i 

- A J (Tr. 56491-93.) 

;l 

25 J 

II 

;i -627- 



(ii) Peripherals . In the early sixties, Honeywell still 
believed that magnetic card devices would be competitively superior 
to magnetic disk drives. It had under development magnetic card 
devices which had been announced to customers. However, the slow 
speed and unreliability of the card devices caused difficulties and 
hurt Honeywell in its marketing of systems. (Withington, Tr. 56494- 
95.) Finally, the effort was dropped, termed in IBM reports "a 
dismal failure". (PX 3481, pp. 75-76.) 

Honeywell made its decision to abandon the magnetic card 
mass storage devices following IBM's announcement of the 2311 disk 
drive for the System/360. Withington testified that this was "a 
major change for Honeywell, because at the time there was no expend! - 



^ '* ture whatever for disk drive development, all of the mass storage 

<3 



.3 

J 



development efforts being put into the magnetic card devices, so 
Honeywell had to start a new effort from scratch and also search 
the industry for OEM sources for suitable disk drives". (Withington, 
Tr. 58562-63.) By 1967 CDC was shipping its 9492 disk file to 



L9 
20 '! 



| Honeywell, who then became its principal customer for CDC 9433 and 

.s : ; 

9434 disk drives, taking in excess of 4,700 units. (G. Brown, Tr. 
51033-34, 51056-57.) Honeywell began to manufacture its own disk 

! packs in 1967, but continued to purchase the drives. (G. Brown, 

21 :i - 

•j Tr. 51056-57; DX 200, p. 31.) 

22 ; | 

Honeywell had already begun efforts to produce itself all 

j of the peripheral devices contained in its EDP systems in the mid- 
24 1 

•j sixties. (DX 13849, p. 28.) During 1965 it started deliveries 
25 i 

■! -628- 



L 
2 

3 

4- 

5' 
a 



of its own card reader, and was about to start shipment of its 
own card punches, (DX 13 349, p. 28.) Prior to 1965 Honeywell 
purchased IBM card, readers and card punches and offered them with 
its own computer systems, including the 200. It planned, however, 
in that time period r to develop its own manufacturing capa- 
bility in punch card equipment, a decision which was accelerated 



7 if by the announcement in late 1964 that IBM would no longer lease 
such equipment to Honeywell and other manufacturers planning 
to re-lease them to customers but would only sell. (Binger, Tr.~ 
4512-13, 4549-50; Spangle, Tr. 5102-07.) 

During 1965 Honeywell introduced new models of printers 
and tape transports and started deliveries of a variety of communica- 
tions terminals as well. (DX 13849, p. 28.) However, it 
continued to acquire software or software development from 
outside companies. (Spangle, Tr. 5092-94.) Contrary to 
IBM's full-scale entry into the manufacture of its own 
components in 1961 (see pp. 282-90 above), Honeywell divested 
itself of its component operations in 1965, stating it "felt 
that we should concentrate our attention on electronic end 

: f products rather than components of this type. We intend to 

20 \ 

i rely on the numerous well qualified suppliers of semiconductor 

21 I . 

% devices for our substantial requirements." (DX 13849, p. 6.)* 

22 j 



3 
9 

ia 

13 
14- 
13 
IS 
17 
IS 
19 1 



23 

| * Honeywell reentered the development and manufacture of 

24. 1 componentry in 1969 with the announcement of a new integrated 

j circuit development center. (DX 123, p. 41.) 

■i -629- 



c. Marketing Practices . Starting in 1965 Honeywell 

offered three- and five— year lease plans for systems and peripherals . 

(Spangle, Tr. 4953;: Brown, Tr. 52613; Withington, Tr. 56624.) 

Mf Under Honeywell's five-year lease plan, the customer could change 

f I his configuration within certain limits without penalty. (Spangle, 

i! Tr. 5529-30.) According to Honeywell's 1965 Annual Report, "the 

j 
T j extended lease terms for our new generation equipment have been 

I j well received by our customers with the result that a substantial 

a. ; percentage of the present leases are being lengthened to five 

i ■ 

j | years. Use of extended lease terms by both present and future 

j 
^l- customers permits computer users to benefit from the new level of 

i 

?\~ technological stability which now exists". (DX 13349 , p. 5.) In 

" ! 

! 
i 

I ';■ 1967 nearly 70 percent of "commercial" Series 200 contracts signed 

i 
^j were for five-year periods. (DX 200, p. 32.) 

Bloch testified that during his tenure at Honeywell, 

c \ (1955-1967) , 80 to 90 percent of Honeywell's computer systems were 

leased. This was dictated "pretty much" by the customer. (Tr. 7673, 

7675-76.) Honeywell developed a sale and leaseback method of 

financing these leases in 1966. Honeywell would arrange with some 

lending organizations — "usually these were syndicates put together by 

someone; like say, the First National City Bank did one group"-- 

whereby it would sell them an amount of installed equipment and then 

lease it back from them. They would give Honeywell the cash for 

the equipment and then it would pay them in installments until it 

had paid back the amount of the cash advance plus a financing 



' ;I 



si 

«i 

i 

'1 ] 

■i 
i 



i 



-630- 



t 

al 

! 



charge. (Spangle, Tr. 5076-77.) According to Spangle, this "improved 
our profit and loss statement" and produced "more cash with which to 
operate". (Id.) 



4*- f ' The sale and leaseback method continued until 1967, when a 

5| wholly-owned subsidiary called Honeywell Finance was set up. That 
S if subsidiary "was able to accomplish much of the benefits of the sales 
7 i and leaseback transaction in so far as creating or attracting cash 
3. I and capital . ► . although it did not have the effect of accelerating 
9- ! the profit from the lease part of it as the sale and leaseback 

IQ i transaction did". It did, however, preserve the residual value of 

I 
H | the equipment for Honeywell. ' (Spangle, Tr. 5082.) Honeywell Finance 

X2. 1 borrowed money from banks and investors through commercial paper 

13 I issuance and through the issuance of long term and medium term 

y±\ bonds, on the security of the receivables from Honeywell's rental 

jjj < contracts. The loan proceeds were passed through to Honeywell. 

i 
IQ ;| (Spangle, Tr. 5082-84.) Honeywell's initial investment in this 

yj ! subsidiary was $15,000,000, half in a subordinated loan and half in 

|q :j common stock. A $60,000,000 line of bank credit was established of 

12 ; which $23,350,000 was being utilized at year end 1967. (DX 200, p. 

20 ! 9 ** 

2t ij During the middle and late 1960s, Honeywell supplied 

22 '■ educational courses for customers, programming support, operating 

I 

23 i| systems and application software without separate charge. (Spangle, 

! 

. A -I Tr. 5084-86.) Bloch testified that this was due to "the dictates of 

:1 
mm ' the marketplace . . . the traditional way in which these services 

63 ii 

j -631- 

;! 

:l 

'i 
•I 



Ij 

i 

a; 
3! 

I 

i 
i 

i 

7 \ 

3 ! 

I 
i 

10 1 

I 

11 ! 

12 J 

13 1 

14,1 

i 

15 ! 

! 

16 ! 
17 
IS 



and equipments were being offered from the time that the field had 
begun". (Tr. 7604.) When IBM announced its unbundling decision in 
1969, Honeywell conducted a study to decide what action to take. 
The study recommended that' Honeywell not follow IBM. Spangle 
testified that there were several reasons for this: 

(1) Honeywell was not set up administratively to handle 
the charge for all these separate items and to enforce their 
collection throughout the field; 

{2h it was not-certain of . the "contractual arrangement" 
it had with its customers, and was concerned that some cus- 
tomers would feel that it had contracted to furnish the items 
which might otherwise be unbundled; 

(3) it also "hoped to gain some temporary market advantage 
. . . because we thought there would be quite a bit of resistance 
to this change by the customers and prospects, and that because 
of that we might be able to get some customers that we otherwise 
would not have been able to get". (Spangle, Tr. 5086-89.) 
Instead of unbundling, Honeywell increased its prices 

19 1 slightly, since it believed that IBM's change would be regarded as a 

20 : ! price increase. (Spangle, Tr. 5089.) Honeywell then began to 

2x ;i advertise its "package pricing" as its "same old bundle of joy . . . 

22 1 once in a while you move ahead just by standing still". (DX 13713.) 

23 \ &" Product and Service Acquisitions and Expansion. 
\ 

?4 j Prior to 1966 Honeywell had developed a series of small, high-speed 

25 ' general purpose digital computers to enhance its capability to 



-632- 



T-3 i 



provide control systems integrated with instrumentation of its own 
manufacture, a related business which Honeywell had been involved in 
for many years. (DX 13849/ P» 20.) In 1966 Honeywell acquired the 
Computer Control Company , which at that time was a leading manufacturer 
of such small, high-performance hardware,* and established it as 
Honeywell's Computer Control Division. (DX 199, p. 3.) The Computer 
Control Company products included the DDP-116, DDP-416 and DDP-516 
computer systems and line of memories. (DX 199, p. 32.) The 
computer systems were used by customers at the time in communi- 
cations switching, engineering and scientific applications. They 
were also offered to OEMs, who built systems for typesetting, 
plotting and freight yard distribution applications. ( Id. ) 

Honeywell applied the "advanced digital techniques" 
, gathered from the acquisition of Computer Control to its own products 
in the industrial process control area (see, e.g., DX 200, p. 24) as 
well as to other areas of data processing. For example, the DDP-516 
was offered for time sharing, communications and medical applications.. 
(DX 7561, pp. 5, 9, 10.) The 516 was made available for use aboard 
ships, aircraft and vans. The modified DDP-516s had all the capa- 
bilities of the standard commercial version — software, price, delivery, 



1 
2 
3 
4- 

3 

« i 

7 

a 

ia 
ii 

12 
13 

15 
16 
17 
IS 

IS ;, 

20. 
•i 

i 

21-1 • 

j * See, e.g., DX 4917, a 1965 letter from John P. Abbadessa, Con- 

22 1 troller of the U.S. Atomic Energy Commission, to a number of computer 
j manufacturers requesting that the AEC's Oak Ridge National Laboratory 

23 • be accorded the benefits of the manufacturers ' educational allowance 
{policy. The recipients were CDC, GE , RCA, Honeywell, IBM, Burrougns , 

2d ! j Sperry Rand, SDS , Philco, Bunker Ramo and Computer Control Company. 

:l 

25 I 

I 

i 

1 -633- 



flexibility and proven design — while meeting the operational re- 
quirements of military, marine and other users. (DX 7727.) Rug- 
gedized DDP-516-s were used by the Coast Guard to gather data for 
weather forecasting. (DX 756 L, p. 10; DX 9087.)* American Airlines 
used 516s to control IBM 1977 terminals within its passenger services 
system* (O'Neill,. Tr. 76000.) The 116 was sold to Bunker Ramo for 
the "control and filing of up-to-the-minute freight booking informa- 
tion" on. airline passenger planes. (DX 5789.) The 116 was also 
used as part of a railroad car classification system by Westinghouse 
Air Brake Company and in process control applications by the Brown 
and Williamson Tobacco Corp. ( Id. , p. 7.) 

In 1969 Honeywell introduced its first expansion of the 
old DDP line: the Honeywell 316. Honeywell called the 316 a 
" minicomputer "** and a "general purpose digital computer". (DX 123, 
pp. 1-2.) The 316 had a full line of peripherals and was offered 
for real-time control, data acquisition and communications applications , 
and as a front end for commercial computers made by others. (DX 7583.} 

Honeywell later incorporated these smaller computers into 
larger computer systems offered by it and offered them for business 



\ 

} 
) 

L ; 

T j 

» 

1 

! 
! 

6 i 

A 
si 

I * The Honeywell press release stated that "using a general purpose 
- j machine rather than specially designed systems formerly employed 

i will let the Coast Guard apply computers to many of its activities 
2 \ at sea". (DX 9087.) We recognize that this press release is not in 

., evidence, but we rely on it both because it represents a contempora- 
2 ■ neous statement by Honeywell describing its products and because the 
facts involved are independently corroborated by DX 7561. 



! ** Spangle testified that a "minicomputer" is a "small general 
15 i purpose computer, and small is a relative term, smaller than other 
1 computers." (Tr. 4916.) 

i 
i 

:i -634- 



I 

2 J 

f 
5 
7 

3 

a 

10' 
11 



data processing. It also sold these same smaller computers to its 
Control Systems Division for incorporation into systems which they 
in turn resold, and to outside buyers for use in specialized systems. 
(Binger, Tr., 4540-42 > 4689*91; Spangle, Tr. 4916-18, 4930.) 

Early in 1968 Honeywell combined its EDP Division and 
Computer Control Division into the Computer and Communications 
Group, to "bring into one organization those related activities that 
are essential to the computer and computer-oriented business". The 
new group was given the mission to involve Honeywell in the "total 
information systems market". (DX 201, p. 31.)* 

In the late sixties Honeywell expanded its EDP services 



J2J. offerings too. It organized an Information Services Division, 



offering "a broad range of integrated remote access computing services 
; and contract software, as well as general data processing and 
continued improved services for Honeywell EDP customers". Sixteen 
data centers were opened around the country. (DX 123, p. 33.) The 
centers used a Honeywell 1648 for time sharing. This system was 
comprised of several Series 16 computers and was introduced by the 



,, competed with the IBM 360/25 and 360/30, the DEC PDP 10 and TSS/8 

1 (based on the PDP 8) , and the Hewlett-Packard 2000 A and B. (Binger, 

21 ;| - 

i Tr. 4593-94.) 

22 i 

r \ 

" 1 

,1 * This Group was Honeywell's contribution to Honeywell Information 

2- i Systems, the newly formed subsidiary of Honeywell which was merged 

J with part of General Electric 's computer operations. (Binger, 
25 ; , Tr. 4531.) 

-635- 



In sum, Honeywell's EDP operations grew steadily throughout 
the 1960s* Honeywell entered the EDP business overseas in 1964 
(DX 132/ p. 11) and its international EDP operations more than 
doubled in 1965. Manufacturing of the 200 series in Scotland also 
began in 1965. (DX I3849> pp. 28-29'.) Its Series 16 was manufactured 
in the Scottish plant and in Japan by its licensee the Nippon Electric 
Ti Company. (DX 201, p. 36.) Honeywell had been rapidly building its 
8 ;| international computer sales force (DX 132, p. 13) ; by 1965 marketing 



Q 

Z 
2 
.4 
5 
.6 



and service organizations existed in Australia, Canada and Western 
Europe (DX 201, p. 36) and by 1969 Honeywell people served 95 percent 
of the "world's computer market" . (DX 132, p. 13.) 

By the end of 1969 Honeywell reported that its Computer 
and Communications Group continued to be the fastest growing area of 
its business; in fact, its computer and communications business was 
growing faster than the industry, and its rate of profitability 
increase was exceeding its growth rate. (DX 123, pp. 7, 28.) 
,- | Revenues for the first nine months of 1969 increased 33.5 percent 
over the comparable period in 1968. (DX 132, p. 11.) Binger stated 
in a speech given in 1969 that domestic computer operations had been 



* 



13 :i 

,_ .{ profitable for four years and overseas operations for two (id; see ! 

M j I 

:} also DX 123, p. 7) ; indeed, the computer business was "solidly profit- ; 

; ! ' I 

;! able". Binger thought that the only possible factor limiting Honeywell's 

— ' i ! 

„- ; { growth was the shortage of qualified people: ; 

23 :i ! 

j "... we are not technology limited, we are not capital ! 

24 ! limited, we are not basically market limited. We may at some ; 

I point be people limited to some extent." (DX 132, p. 17; see 

-- .j Binger, Tr. 4818-20.) i 



;| -636< 



L 
2 
3 



In 1969 the Computer and Communications Group employed 24,000 people 
worldwide, compared with 18,000 a year before. Over one-half of 
Honeywell's research, development and engineering dollars were spent 
in the computing area. (DX 123, p. 33.) Between 1963 and 1969 
Honeywell's domestic EDP revenues had increased from $27 million to 
$210.8 million, a more than sevenfold increase in seven years. (DX 



3 

j . jj 8224, p. 387; DX 14484, p. 1? DX 8631, pp. 31, 37.) 
3 ! 

lo ; 

u 

12 

13 

14 i 



16 3 



17 j 
13 
13 
2Q I 



22 :s 

t 
2*1 

■i 
•t 

?* i 



-637- 



y 

i 

i 

l 

c i 



44. Burroughs . The story of Burroughs in the period 1964 
to 1970 is one of a company that turned itself around, going from 
predictions . of failure to success . A slow starter in computers , by 
1964 Burroughs had still not 'transformed itself into much of a com- 
puter company, having developed and marketed relatively few EDP 
products, (See above, pp. 227-28 •) Beginning in 1964 Burroughs 
shook up its operations, reduced expenses and, while remaining profit- 
able, increased its investments in research and development. The 
results were a proliferation of new products , substantial growth and 
increased profitability over the decade. 

a. Burroughs in. 1964; Problems and Changes. Burroughs 1 
situation in 1964 did not look promising for future growth in the 
computer industry. Indeed, as R. W. Macdonald* wrote in 1975: 



"[I]n 1964, some analysts who observed the developing computer 
industry, had serious doubts about the ability of Burroughs to 
survive in the new environment as a computer company. Even some 
members of our own Board of Directors were concerned, and a 
ff it highly respected financial journal predicted flatly that Burroughs 

i! either would have to merge into another company or fail . " (DX 

it 427, p. 2.) 

1 

j Those serious doubts were based on two factors : Burroughs ' mediocre 

S '■! 

;} record in computers and the perceived strength of its competition. 



si 



Burroughs' record in computers as of 1964 was not strong. 



\ * R. W. Macdonald, a director of Burroughs since 1959, in 1964 
— j became Executive Vice President, in 1966 became President and Chief 

■j Operating Officer and in 1967 became President and Chief Executive 
2 j Officer. (Tr. 6882-83.) Macdonald testified at trial by deposition. 



:? 



-638- 



L 
Z 
3 

*i 

f 

« ! 
7 
3 
9 

ia 

LI 
12 

13 



It had begun., but had not completed, the "major transformation" from 
mechanical office equipment to electronic computer technology. 
(DX 427, pp. 3-4 r see above, pp. 227-28.) Its computer product 
line was limited, consisting mainly of the B 200, B 5000 and D 825 

computers* (See above, p. 227-) its financial record since 

1961 had been poor; its revenues had "remained on a plateau" and its 

earnings were "unsatisfactory". (DX 427, p. 3.) 

The doubts about Burroughs ' future were also based on the 
"size, profitability and technical achievements" of the "many compa- 
nies [who] had aspirations to be mainframe manufacturers". Regarding 
the size of those companies, Macdonald wrote: 



"We faced giants such as RCA, with 1964 revenues of over 
$2 billion; Honeywell, with over $600 million; Sperry Rand, 
with its Univac Division, with $1.3 billion. IBM in those days 
had revenues of over $3 billion, but IBM was not the largest 
* . jj company we faced in terms of total revenues. General Electric, 
^ ; with serious intentions and a major program in computers, 
already was an industrial giant with revenues in excess of 
$5 billion." (DX 427, p. 3.) 

10 l| In contrast Burroughs' total annual worldwide revenues were less 

- 7 |l than $400 million. (DX 10260, p. 22.) 

-^ lj With regard to technical achievements, Macdonald wrote 

*►- S that by 1964 IBM was "well on their way to development of a truly 

impressive research and development capability" and "General 






21 :! Electric had been exploring the uses of the electron for years in 
22 '1 both electrical and electronic applications". By contrast: 



23 



"Although [Burroughs] had been engaged in electronic research 
and had achieved initial success with a few very advanced 
24 i new products , the products on which our revenue and profits 
depended remained primarily mechanical." (DX 427, p. 3.) 



-639- 



1 
z 

3 



5 

a 

7 | 
8-! 
9- 

10 

11 



Starting in 1964 and continuing through the 195 0s 
Burroughs set about to achieve its objective of "profitable growth" 
and "moderate growth commensurate with maintaining profitability" in 



4* j computers . (Withington, Tr, 56732; DX 10262, p. 6.) As a first 



step, in 1964 Burroughs' President Ray Eppert formed the Profit 
Improvement Committee. The Committee was to consider reorganization 
"with respect to all aspects of marketing, manufacturing and engineer- 
ing operations, and the establishment of clear product development 
objectives". Its "primary charge was the swift improvement of the 
company's profitability". (DX 427, p. 5.) 

The changes instituted by this Committee (of which 



75 I Macdonald was a member) and further changes instituted by Macdonald, 



who, in 1964 was given "broad administrative responsibilities" (DX 

10260, p. 4) and in 1966 became Chief Operating Officer (Tr. 6883), 

were intended to accomplish two things: first, reduced expenses, and 

second, improved development of computer products. (Macdonald, Tr. 

If 6883-91; DX 427, pp. 7-8.) 

i| (i) Reduction of Expenses. The Profit Improvement Committee 

i 

ij found that Burroughs ' " * problems ' lay in the efficiency of its 

| operations" and not in "spending levels associated with research 

i 

■I 

•j and development". The Committee instituted several changes* to 



13 i 

i 

> 

14*! 

15 

16 

17 

IS 

19 * 

20 

21 ;; 

22 ;i 

„ ij * These changes contrast with the policies implemented at RCA 

25 i during the late 1960s. At RCA expenses increased, the importance of 

j financial controls was not emphasized, product development was not 
2"" ij encouraged, and market share rather than profitability was considered 

•! the goal. (See pp. 531-89 above.) 

1 

ii 

■:l -640- 



L 
Z 

3 
4t 
3 
S 

7 
3 
9 

ia 

II 

12 

13 
14 
15 
IS 
1? 
IS 
19 



increase efficiency: 

First , the productivity of the sales force was increased. 

To do this, the Committee reduced salaries and commissions for 

salesmen, reorganized the sales organization* and moved unpro~ 

ductive salesmen out of the division. Burroughs found that: 

*The combined effect of these organizational changes 
gave us the equivalent of adding 500 highly productive 
salesmen— with no increase in budget costs." (DX 427, 
p* 7.)** 

Second , the Committee found that manufacturing costs 
"had been increasing as a percentage of revenue every year for ten 
years" , and the Committee undertook to reduce those costs . 
It did this by reducing the number of managers at its plants, 
specializing the plants by products, introducing a series 
of financial controls, designating each marketing district 
a "profit center", modernizing existing facilities and build- 
ing 17 new plants. With these changes, by 1966 manufacturing 
costs were reduced by more than five percent of revenue and 
continued downward in ensuing years. (DX 427, pp. 7-8.) 
Those reductions of expenses soon benefited Burroughs: 



20 > * Burroughs created additional sales offices, established sales 

;| zones, reduced the number of salesmen reporting to each manager, and 

21 |j gave each manager a personal sales territory to cover. (DX 427, pp. 
ii 6-7.) 

22|! 

i| ** In its 1965 Annual Report Burroughs reported that its "market- 
23 | ing realignment program [had] contributed to the sales success of 

:[ Burroughs' business machines and systems and improved profitability 
24. of the Corporation in 1965". (DX 10261, p. 8.) 

25 \ 

-641- 



L 

2 

Z\ 
\ 

3 

a; 

7 

8 

9- 
10 
11 
12! 

14 
15 ! 
16 

V I 

18 I 

i 

i 

19 i 

20 
21 
22 i 

23! 
24! 
25 ! 



"The combined effect of these major changes, along with 
reductions in marketing and G&A expenses and other economies 
resulting from stricter overall control, produced an increase 
in net earnings of over 2.00 percent in two years, from $10 
million in 1964, to $31 million in 1966." (DX 427, p. 8.) 

(ii) Increased Product Development. As a threshold 

matter the Profit improvement Committee decided that Burroughs' lack 

of profitability did not result from too much spending for research 

and development. (DX 427, p. 6.) In fact, the reduction of expenses 

discussed above allowed increased expenditures in research and 

development : 

"Removal of these excesses resulted in greater profit- 
ability, which in turn made more money available for research 
and development, allowing us to spend more in engineering, 
leading to further cost reductions. Since 1964, we have 
increased our commitment to R and D each year. ..." (DX 427, 
p. 8.) 

In 1964 Macdonald began to have "greater influence" in 

Burroughs and his "principal activity was to utilize these resources 

and developments to a much greater degree than they had in the 

past" . In addition, he made sure that Burroughs would "pay a great 

deal of attention to product development". (Macdonald, Tr. 6886.) 

Burroughs pressed ahead with its computer developments in 

two ways: First, it expanded its "product program to become more of 

ja full range company". (Macdonald, Tr. 6888-89.) Second, it offered 

; greater capability and increased the diversity of its computer 

products. (Macdonald, Tr. 6889-90.) The addition of new products, 

in turn, made more money available for research and development, as 

Macdonald explained: 



-642- 



1 

2 
3 
4 
5 

6 

7 
3 

* 
10 

u! 

I 

12 | 

-I 

I4| 

15! 

16 

17 

IS 

IS 

20 

21 

22 

23 

24 

25 



"In 1964 we were operating on a research budget, an R and 
D budget of approximately sixteen million dollars and as we 
expanded our business we were able to afford an expanded R and 
D budget and as we were able to do this we expanded the range 
of products which we felt we could successfully undertake . . . 
CTr. 6889.) 

By 1969 Burroughs' annual spending in research and develop- 
ment had doubled to $35 million. (DX 10285, p. 3.) 

The changes that Burroughs began in the mid-1960s, par- 
ticularly its increased research and development and improved manu- 
facturing capabilities, required new investment. Macdonald described 
those investments between 19^5 and 1972: 

"Since 1965, Burroughs had spent some $250 million in R&D. 
These funds came entirely from our own resources and were used 
for the development of our commercial and trademark product 
line. 

"Over the same period, we have also invested just over one 
billion dollars to expand the manufacturing and marketing 
facilities to sell the products resulting from this R&D expen- 
diture. Approximately $750 million of this represented a 
marketing investment. It went for facilities, inventory, 
receivables and lease funding. The remaining $250 million was 
for manufacturing facilities, men, machinery and equipment. I 
should also point out that this billion dollars was in addition 
to the $500 million that we had already invested by the end of 
1965. Of the billion dollars invested over the last seven 
years, $250 million was generated through retained earnings and 
I) the remaining $750 million was raised in the financial markets 
through loans and equity issues." (DX 426, pp. 19-20.) 

b. Computer Developments 1964 - 1969. Burroughs moved to 

extend both the breadth of its product line and to increase the 

capabilities offered by its computer products and by the end of 1969 

had succeeded in adding many new products. This discussion traces 

j the development of Burroughs ' computer products in three parts : its 

500 Systems, its smaller computers and its peripheral products. 



-643- 



(i) The 500 Systems Family. An important factor in 
Z j Burroughs ' success during the 1960s was the success of its 500 

Systems family. Nine systems in that family were eventually announced: 
4».| the B" 500,* B 250:0, B: 3500, ,.B 4500, B 5500, B 6500, B 7500, B 8300** 
; and B '85 0O- Because of problems that Burroughs, in common with 
other manufacturers, experienced with its larger machines, the B 
7500, B 8300, B 8500 were either not delivered or not operational at 



q J customer locations, and the B 6500 was delivered late. The B 4500 was 
also never delivered. (PX 5048-D (DX 14506), Pierce, p. 62.) Still, 
by 1969 Burroughs, was able to report that "this family of balanced 
general purpose commercial data processors have helped the Corporation 



in.! 
ii 



13 



^ j establish an excellent position in the EDP market". (DX 10264, p. 5.) 



However, while Burroughs promoted the 500 Systems as a "family", they 



. . ; were not machine-language compatible as was the IBM 360, but were only 



compatible through the use of higher level languages. (PX 5525-A, 
p. 218; DX 10264, pp. 6, 8.) 

Four months after IBM announced its System/360, Burroughs 
in August 1964 announced the first member of the "500" System family, 
the B 5500. (DX 13920.)/ Burroughs described the B 5500 as a 



15 
IS 

13 ! 

19 I 

20 \ 

II * Burroughs sometimes promoted the 3 500 as part of the 500 
2* |j Systems (DX 10264, p. 6); it was, however, more closely related to 
'•I the B 200 and B 300 and is discussed in that section. 

22 !■ ' 

I ** Burroughs did not describe the B 8300 as a member of the 500 

22 ! Systems family. However, it was closely related to the B 8500 and 

is therefore discussed in this section. 
24 

/ Withington commented that, by announcing only one new model, 

25 , Burroughs had not "attempted to answer Svstem/360 across the board". 

(PX 4829, p. 22.) " | 

i 
i 

-644- I 



i 

5 | 

si 

i 

r 

i 

I 



"modular data processing system of advanced design for both com- 
mercial and scientific applications in the medium to large scale 
categories". (DX 10260, p. 12.) Burroughs reported that it had "up 
to three times more productivity than its predecessor, the B 5000" ► 
(Id.) 

B 5500s were indeed used for commercial and scientific 
applications, as well as in aid of the space program. During 1964 a 
B 5500 "joined the famous Atlas ground guidance computer in the 
nation's space program" and was used to track the Saturn missile. 
Two Burroughs' B 5500s were also used in "tabulating and projecting 



l± ! national election results for the American Broadcasting Company". 



Those B 5500 Systems "operated in the same manner for the election 
as all Burroughs' computers do in projecting business trends, statis- 
tics, competitors' activities and other information on which manage- 
ments make decisions". ( Id. ) By 1965 Burroughs reported that its 
orders for the B 5500 had exceeded forecast and "included many 
diverse applications in national and state governments, advertising, 
manufacturing, shipbuilding and research". (DX 10261, p. 9.) 

During 1965 Burroughs' Defense, Space and Special Systems 
Group* announced the B 8500. The B 8500 was marketed for "high 



12 

13 

L4~ 

15 

IS 

17 

IS 

19 

2G 

21 

22 

23 

24 'i 

: "The Defense, Space and Special Systems Group produces 

?5 ij and markets special data processing systems and advanced products 

for the military and other government agencies. It manufactures 

-645- 



* Burroughs had two groups that marketed computers, its Business 
Machines and Defense, Space and Special* Systems Groups. In 1968 
Burroughs described the functions of its Defense, Space and Special 
Systems Group: 



volume, time- sharing, on-line business , scientific and government 
applications" and provided for "management information processing, 
including the full complement of business data processing, reporting 
and. message handling as weXl as centralized or decentralized scien- 



super-computer systems for government, commercial, educational 
and scientific applications * The Group also produces visual 
display systems , memory systems and electronic components . " 
(DX 10263, p, 17.) 

The relationship between the Defense , Space and Special Systems 
Group and Burroughs' commercial business was close and involved, 
marketing and designing the same or similar products. Burroughs 
described the relationship in its 19 6 4 Annual Report r 

" [tlhe Corporation's programs for various military and civilian 
agencies, coupled with large investments of its own in research 
and development, have yielded important technological advances 
which are being utilized in Burroughs ' commercial data processing 
systems and accounting machines as well as in defense and space 
projects". (DX 10260, p. 19.) 

Similarly, in its 1965 Annual Report, it stated that: 

"[t]he Defense and Space Group was expanded in 1965 to include 
the development, production and marketing of custom-built 
large-scale electronic data processing systems for commercial, 
industrial and special applications . These systems to a great 
degree now parallel the requirements of high performance com- 
puters employed in major defense and space programs where 
Burroughs has had many years of successful experience." (DX 
:j 10261, p. 17.) 

I Products designed and marketed commercially often were later 
1 J modified or further developed and marketed for military use, and 

i vice versa. For example, the B 5000, the foundation for the subsequent 
I j "500" product line, grew out of military work (the Burroughs D-825) . 

3 (Withington, Tr. 55976-77, 58527-28.) And when the Defense, Space 
1 '\ and Special Systems Group was awarded a contract to produce a mobile 

i| communications system for use by the U.S. Army, it modified four B 

3 : i 3500 computer systems which had been developed by the Business 

[Machines Group. (DX 10716, p. 8; DX 13665, p. 19.) 
4, 'j 

■! At the end of 1968 the Defense, Space and Special Systems Group 
5 j took over the responsibility of marketing, in addition to specially 
I designed equipment, all of Burroughs computer products to the Federal 
'government. (DX 13665, p. 19.) 

i 

;i 

i -646- 



L 
2 

3 

4 



5 } ; monolithic integrated circuits . (DX 10261, p. 17.) 



S 
7 

Id 

II ! 

12 i 

13 
14 
IS 
15 



tific and engineering computations". According to Burroughs, the 
B 8500 was a "logical extension" of the 'concepts of "modularity, 
multiprocessing and automatic scheduling programs used with the 
B 5500 and D 800 series systems..* The B 8500 also made use of 



By 1967 Burroughs reported that: 

"Broadening customer interest in the giant self -regulating B 8500 
system confirms the importance Burroughs has given the develop- 
ment and production of this supercomputer. It has the unique 
ability to multiprocess a number of batches of accounting 
routines, solve engineering and scientific problems, and deal 
with transactions as they occur, all at the same time. The 
interest of potential users in the B 8500 has greatly increased 
for on-line, real— time business and scientific applications." 
(DX 10263, p. 9.) 

However, Burroughs experienced problems developing the 
B 8500, and none was ever delivered. (Perils, Tr. 2001-02; PX 504&-D 
(DX 14506), Pierce, p. 62.) 

It was not until 1966 that Burroughs began to turn its 500 
Systems into a family of computer systems somewhat comparable in 



is ; 

19 
20 



t7 I breadth to IBM's System/360. In that year Burroughs introduced 



three new "members of the 500 systems", the B 6500, the 3 2500 and 
B 3500. (DX 10262, p. 8.) 

Burroughs reported that the B 6500 was "taking [its] 



21 ij place" between the "medium- sized" B 5500 and the "giant" B 8500 



22 



-„ i| * The D 800 series included the D 825, a computer developed for 
" ; the military. (See above, p. 227.) Macdonald testified that 



24 

25 



a good deal of the B 8500s "architectural concept came from the 825" 
and it "was intended to be an enlargement and in terms of size and 
speed from the generation of equipment which was the D825". (Tr. 
7556-57.) 

-647- 



3* 

Si 
7 

a 

9- 
10 
11 : 
12 
13 
U! 
15 

17 ! 

18 J 

19 i 

20 I 



L I The B 6500 central processors employed monolithic integrated circuitry 
Z j throughout; had core or thin— film main memories; were "equipped for 
Z-\ true multiprocessing, parallel processing, and real-time and time- 
4- 1 sharing operations" ,. and had a "comprehensive , automatic operating 
system for program control, completely coordinated with the hardware 
elements". (DX 10262, p.. 8.) 

The B 6500 was not delivered until 1969. (DX 10264, p. 8.) 
Even, then its "full development" was delayed by problems in its 
system software. (DX 3269, p. 3.) Burroughs reported that it had 
corrected those problems in 1971. ( Id. ) 

The B 2500 and B 3500 were released for sale in April 
1966. (DX 10262, p. 3.) Demonstrations for these systems "were 
made on a broad range of business applications programmed in COBOL, 
including remote processing and multiprocessing under the automatic 
control of the Master Control Program" . ( Id. ) Burroughs reported 
these systems would be sold "in the medium-priced range". ( Id. , 
p. 8.) By 1967 Burroughs reported that it had received "an impressive 
number of orders" for the B 2500 and B 3500 from users in "such 
diverse fields as finance, manufacturing, government, retailing, 
insurance and publishing". (DX 10263, p. 9.) 
-, ;| In 1967 Burroughs announced the B 7500. Burroughs reported 

;! that its release "stimulated interest in other EDP products and 
; strengthened the Company's position in this highly competitive 
field". (DX 10263, p. 11.) However, the B 7500 was never delivered. 
(PX 5048-D (DX 14506), Pierce, p. 62.) 



24 
25 



-648- 



L 
Z 
Z 



In its 1968 Annual Report Burroughs reported that during 
1968 it had installed the B 8300, "part of the B 8500 development 
.program;" , to provide "a central passenger reservation system for a 



4^| major world airline". That installation used "three central pro- 
£ !t cessors functioning under the automatic control of a single software 
5 '| operating system"; there were more than 2700 input and display 
j 1} terminals throughout the United States with keyboard input and 



8 
3 



IC l 



II 

121 

13 

14. 

13 



10 



cathode ray tubes "to display data transmitted to and received from 
the computer". (DX 13665, pp.. 3, 5.) 

The airline at which Burroughs installed its B 8300s was 
Trans World Airways (TWA). (O'Neill, Tr. 76014.) The B 8300s at 
TWA were never operational, however, because the B 8300s "could not 
• accommodate the projected workload" and Burroughs "had not demonstrated 
adequate availability or reliability of the system". (O'Neill, Tr. 
76015.) The effort was terminated, and in late 1970 TWA sued 
Burroughs for non-delivery of the B 8300. (O'Neill, Tr. 76015.)* 
In 1971 TWA installed one IBM 360/75 and two IBM 360/65s to 



17 ., 

perform the reservations function.** (O'Neill, Tr. 76013-14.) 



19 : 

';' * The litigation was settled in October 1972 with Burroughs 

20 ; ! agreeing to assume certain payments to a leasing company and either 
ij to make equipment available or to pay a sum of money to TWA. As a 

21 j result of the settlement, Burroughs' earnings were reduced by 
:j $4,813,000 net of taxes. (DX 10265, p. 42.) 

22 :i 

'ij ** Burroughs was not the only company that had difficulty installing 
23 an airline passenger reservation system during the 1960s. Sperry 

! Rand was also unsuccessful at installing such a system at United 
24, 1 Airlines. United, like TWA, then acquired IBM equipment. (O'Neill, 

JTr. 76015-17, 76231-32.) 

25] 

i 

'! -649- 



By 1969 Burroughs reported that the production of its 

"'500 Systems' reached an all-time high during the year". Burroughs 

described some of the reasons for the success of the 500 Systems: 

"Our systems software provides self -regulated operation 
which assures Burroughs customers of maximum work output through 
the techniques of multiprogramming in which a number of different 
programs are handled at one time. In the larger systems, 
simultaneous parallel processing of programs is achieved by use 
of multiple central processors. Another important advantage to 
users of our medium and large systems is a modular architecture 
which enables them to add processors and increase main memory 
and input/output capacity in increments as needs expand. 
Upward compatibility— from one '500' Systems computer to the 
next largest in size — is assured through the use of higher 
level programming languages. COBOL is used for business appli- 
cations on the entire range of computers. For engineering and 
scientific applications, FORTRAN is used on the medium-scale 
B 3500 and B 5500 and the large-scale B 6500 and B 8500, and 
ALGOL on the B 5500-, B 6500 and B 8500. PL 1 will be available 
on the large systems and other special languages will be added 
as they are required." (DX 10264, pp. 6-8.)* 

These characteristics, of course, were much the same as 

those IBM had earlier employed successfully in its System/360. 

(ii) Smaller Computers. Burroughs marketed its smaller 

;l computers in three lines: its line of B 200 successors, its E 

j ' Series and the L/TC Line. 

Burroughs had introduced the B 200 in 1961. (PX 552 5A, 

p. 53; see above, p. 227.) In early 1965 Burroughs introduced 



a : 
s 



2 :i 

3 



'! 



* Burroughs reported that it had been able to obtain a design 



Sj advantage with the Burroughs 500 systems: 



"[w]ith the Burroughs 500 Systems, the corporation gained an 
advantage by developing the software and hardware in parallel. 
Engineers in these two areas combined their efforts as the 
systems were developed, closing the time lag between installation 
and complete usefulness of the system to the customer. This 
advantage also insures the user maximum performance of the 
complete system." (DX 10262, p. 10.) 

-650- 



L ; 
2; 

i 

3 ! 



4 i one B 300 processor with a single disk file system". (DX 10260, 



5! 
i 

! 

6 : 



s 

9 

10 
11 
12 
13 

14- ;t 
15 

is ; 



its B 300 data processing system which was compatible with 3 200. 
Burroughs reported that the B 300 included on-line capacity and that 
its "modular design provides for the simultaneous use of more than 



p. 12.) By 1966 Burroughs reported that the B 200 and B 300 computers 
had been: 



j i " [1] eased or purchased by customers in many fields including 
transportation, data processing services, photo supplies, 
utilities, insurance, publishing, brewing, school systems, 
manufacturing, baking, textile milling, property management, 
retailing, wholesaling, distributing, government and public 
service, research and finance." (DX 10262, p. 12.) 



In 1965 Burroughs introduced the B 340 bank data processing 
system which was smaller than, although compatible with, the B 300 
system. (DX 10263, p. 11.) During 1968 Burroughs introduced its 
3 500 computer. The B 500 had an automatic operating system and used 
COBOL. While promoted by Burroughs as a "member of the '500' Systems 
EDP family", the 500 was compatible in assembly language with the B 
100, B 200 and B 300 systems but was compatible only through the use 
*7 i| of COBOL with the 500 Systems. (DX 13665, p. 5.) 

1 a I 

*° ;i During 1964 Burroughs brought out its E 2100 computer. 

19 1 

^ !j (DX 10260, p. 8.) Between 1964 and 1970 it added to the E Series 

20 j with the E 3000, E 5000, E 6000 and E 8000. (DX 10263, p. 11; 

2- ! j DX 10264, p. 18.) The E Series were small solid state computers 

jl 

22 j with electronic logic and data storage. • (DX 10260, p. 9; DX 10264, 

22 -j p. 18.) On the larger E Series computers, the E 6000 and E 8000, 

; t 

24 :j COBOL was available. (DX 10264, p. 18.) 

In 1968 Burroughs took a major step forward with the 

-651- 



25 



L 
2 
3 

4 

5 

5 

7 

3 

9- 
10 
11 
12 ! 
13 
14 
15 
IS 
17 
IS 
IS 
20 
21 
22 
23 
24 



announcement of its TC 500 terminal computer. (DX 13665, pp. 1, 7.) 
The TC 500 was characterized by John Jones of Southern Railway as 
the first "intelligent terminal", that is, the "first programmable 
terminal . . . that had in it a processor, a general purpose pro- 
cessor with memory and input and output, that could be programmed to 
perform in some way as the user desired as opposed to being hard 
wired". (Tr. 79044-45.) The TC 500: 

" [h] ad a keyboard for an operator to input data and a printer 
on which data could be printed, a character printer, and a 
processor inside of it which could be programmed to give that 
device any particular characteristics in its. operation, as well 
as do other processing of the data as it was entered or before 
it was printed." (Tr. 79044.) 

Burroughs similarly described the TC 500 in its Annual Report:. 

"An internally programmed computer using integrated circuitry 
and dis3c memory, the TC 500 can operate as a data communications 
terminal on-line to a central computer, or function off-line 
independently. ... In addition to data communications, 
they can edit and format information and perform functions 
which previously had to be handled by the central computer." 
(DX 13665, p. 7.) 



In 1969 Burroughs introduced its TC 700 and TC 310 terminal 
I 
} computers. The "TC 500 and TC 700 have their own computing and 

j memory capabilities. They also edit and format information for most 

| economical transmission to a central system. The TC 310, in multiples, 

j is connected to a data controller which then performs the formatting 

; and other necessary operations prior to data transmission." (DX 

'' 10264, p. 14.) 

Also resulting from the same engineering as the TC computers 

was the L 2000 computer. Introduced in 1969, the L 2000 was a 
\ 

^ ; computer, designed for billing, as to which "the addition of a data 



-652- 



L 
2 

3 

5 
S 
7 
3 

10 

11 j 

i 

13 i 

i 

14- I 
i 
i 

is ; 
is I 
17 ! 

13 ■ 



communications unit converts it to a terminal computer able to 

communicate with a central computer system". (DX 10264, p. 18.) 

COBOL was available for the L 2000. ( Id. ) 

Macdonald described the L/TC series as follows: 

"These internally programmed machines are programmed in COBOL 
and can operate under operator control or under program control. 

ti 

• . • . 

"These small systems are, in terms of what they can perform, 
small full-scale computers." (DX 10285, p. 6.) 

(iii) Peripherals. During the years from 1964 through 
1969 Burroughs improved upon its existing peripheral equipment. It 
introduced several models of improved card readers, printers, sorter- 
readers, tape transports, multi-tape listers, and tape drives. (DX 
10261, p. 11; DX 10264, p. 10.) 

By 1964 Burroughs had developed and was marketing a disk 
file with a head-per-track. (Withington, Tr. 56244; DX 10260, p. 10.) 
This head-per-track file had a slightly faster access time and a 
slightly higher cost per unit of storage than the movable head 
devices. (Withington, Tr. 56244-45.) During the mid-1960s Burroughs 
'! found that its disk drive was "a significant factor in the growth of 



19 

2Q 1 the Company's business in EDP systems". (DX 13665, p. 5; see also 

-, ":| PX 4834, p. 31.) 

22 l! However, in 1962 when IBM introduced its 1311 disk drive 

si 

-- ! | with a removable disk pack, Burroughs did not offer a disk drive 
- A ;[ with a similar removable pack, nor did Burroughs offer such a disk 
„ g j drive after IBM followed this announcement with the introduction of 

-653- 



i 

t! 

3 



3 

a 
z 

■** 
-5 

1 «r 

.3 

L7 



LS;j 



the 2311 and 2314 disk drives. Where Burroughs' customers wanted 
the advantages of a removable disk pack, Burroughs sought to convince 
them to keep their files on magnetic tape and to load and unload the 
files on to the Burroughs' fixed pack drives. (Withington, Tr. 
58802.) Finally, in the late 1960s, Burroughs arranged to acquire 
5 ;' disk drives with removable disk packs from Century Data, and in 1970 
7 : ;j it began marketing those disk drives as part of its computer systems . 
(PX 4445, pp. 7-8; DX 10716, p. 12.) 

Burroughs introduced new peripheral equipment during the 
1960s. In 1969 Burroughs introduced a new electronic reader/sorter 
which handled documents both optically and magnetically encoded and a 
new computer-output-to-microfilm system. It also announced three 
new encoding devices : the Series N keyboard-to-magnetic tape data 
encoding machine; the A 149 peripheral card punch, the A 150 keypunch, 
and the A 160 verifier for punched card encoding; and the Series S 
"general purpose character encoding machines . . . designed to 
encode unit documents ... to facilitate electronic reading by 
high-speed recognition equipment". (DX 10264, pp. 10-14.) 

c. Burroughs at the End of the 1960s. By the end of the 

I decade the changes Burroughs had instituted in 1964 had begun to j 
ZO "! | 

i achieve Burroughs' objective of "profitable growth". (DX 427, p. 8.) j 
21 1 I 

:j Burroughs had reduced costs and increased efficiency in its manuf ac- j 

^ 1 ! 

:l turing and marketing operations (DX 427, pp. 6-8), and it had increased j 

23 ■ 

jits expenditures in research and development. (Macdonald, Tr. 6 889; j 
2- i I 

! DX 427, p. 8; DX 10264, pp. 4, 6.) Aided by those changes, Burroughs j 

1* \ 



-654- 



L 
2 
3 

3 

6 
7 
3 
9 

ia 
ii 

12 

13 

14; 
15 
16 II 



had expanded its product line in terms of both range and the capa- 
bilities offered. From a few mid-size computers in 1964, Burroughs 
delivered several complete lines of computer systems ranging from 
small (E Series; L/TC Series) to very large (B 6500) by 1969. Its 
numerous new product offerings were reflected in its 1969 Annual 
Report which described its "broad line of products for the data 
recording, computing and processing market" including: 

"[c]omputer systems, memory sub-systems, peripheral 
input and output equipment, data encoding equipment, data 
communications terminals, accounting systems, calculators and 
adding machines, business forms and office supplies, custom- 
designed electronic systems, and data display devices. This 
extensive range of products represents one of Burroughs basic 
strengths for continued growth in the rapidly expanding data 
processing industry." (DX 10264, p. 2.) 



And Burroughs was also continuing its technological development in 
intelligent terminals, an area that would become very important in 
the 1970s. (See DX 10264, p. 14.) 

It was clear that Burroughs ' management understood the close 
interrelationship of its extensive product line. In a 1969 presenta- 
^' ij tion to the New York Society of Security Analysts, Ray Macdonald 

stated, concerning the relationship among various computer products, 



13 



19 ;j 



,j that: 

^ J .| "In 1967, I said that when I had the next opportunity of • 

•j addressing this group we might refer to electronic accounting 

2- !! machines, electronic accounting systems, terminal units and 

•I electronic computers as one continuous market from small machine 

2~ jj to giant computer. This blending of several markets into a 

__ i( single broad market has now become more evident." (DX 10285, 

^ :| P. 5.) 

** ' Burroughs' financial results, in turn, reflected the 

25 ; proliferation of its computer products. From 1964 to 1969 Burroughs' 

-655- 



total corporate revenues did not quite double, increasing from $392 

million to $759 million. During the same time its domestic EDP 

revenues increased from $61 million to $260 million and its corporate 

» 

profits jumped 500 percent. (DX 8224, p. 1; DX 10260, p. 5; DX 10264, 
5 I P. 5.) 

Writing in 1975, Macdonald looked back on the results of 
7 | the changes that Burroughs had instituted in 1964: 



3 

ixi 

i 

12.1 

13 ! 
i 

14.! 
15 

16 ;! 

17 j 
IS ! 

13 I 
20 ; 



21 || 



22 -j 

23 j 

24 :| 



"Our revenue has doubled every five years, and today, at 
$1.5 billion, is four times its level of ten years ago. 

"Our net earnings have increased by 14 times during 
the 10-year period, and this is the best record of growth in 
the mainframe computer industry. 

"Our manpower worldwide has increased from about 34,000 
to more than 51,500. We are operating 54 plants in ten countries 
and two more plants are under construction." (DX 427, pp. 2-3.) 

By the beginning of 1970 Burroughs had made up much of the 

ground it had lost during the 1950s and early 1960s, and was well 

situated for even greater success in the 1970s. 



-656- 



1 45. National Cash Register. Historically, National Cash 

2 [Register (NCR) was a company that had concentrated on marketing its 

i 

3 ! "traditional products" — cash registers, accounting machines and adding 

4 | machines — to customers engaged in retailing and banking. (DX 344, p. 

5 ;1; DX 372, p. 1; see pp. 229-31, 236 above.) By the beginning of 

6 11964, while continuing to concentrate on customers in those areas, 

7 NCR had introduced and was marketing two models of its second 

8 generation 351 computer system which had been announced in the early 

9 1960s. (DX 344, p. 14; DX 382, pp. 3, 10.) At the same time NCR 

10 ! was actively expanding "the functions of its traditional products". 

i 

11 (See above p. 241; DX 344, p. 1.) In 1964 NCR's domestic EDP revenues 

12 ($46.3 million) accounted for only about 13 percent of its total 

i 

13 domestic revenues. (DX 361, p. 22; DX 8224, p. 3.) 

j 

14 | The story of NCR in the years between 1964 and 1970 is that 
i 

15 of a company wishing to maintain its traditional business and only 

16 I gradually adding the increased capability offered by computers. This 

17 i| desire gradually to develop computers to support its traditional 

ii 
ii 

IS j business was expressed by the President of NCR, Robert S. Oelman,* 

19 j in a November 1964 speech. He stated that the company had "recently" 

20 undergone "the most significant change in [its] long history . . . the 

21 advent of electronic data processing." (DX 342, pp. 2-3.) However, 

22 

23 j * Oelman, along with J. J. Hangen, were the two witnesses called by 
j plaintiff from NCR. From 1964 until he left NCR in 1973, Oelman was 

24 j Chairman and Chief Executive Officer of NCR (Tr. 6117) , and from 1964 
! until 1972 Hangen was Vice President of Finance. (Tr. 6233.) At 

25 i the time he testified Hangen was Senior Vice President of Corporate 
Affairs for NCR. (Tr. 6239.) 

-657- 



1 ! 
I 
i 

2 ! 
3J 

i 

5 I 

i 

s ! 

7 | 

3 ! 

i 

9 ! 

i 

111 

12 | 
13j 

i 

14.! 

.1 

i 

15 i 

t 

I 

16 : 

17 !i 



Oelman explained that this change did not mean the demise of NCR's 
"traditional products" — cash registers, accounting machines and adding 
machines — for two reasons: First, the traditional products were 
"being integrated" into electronic data processing systems,- the tradi- 
tional products serve as an "input medium" for data and are tied into 
"the mainstream of the data processing revolution". Second, NCR could 
use "new technologies to add important machine features and to improve 
overall performance" of its traditional products. ( Id. , pp. 3-6.) 

Thus, NCR, rather than recognizing that computers were going 
to obsolete its "traditional business" (as IBM had in the 1950s) and 
committing itself to the new technology, chose to split its resources 
between computers and its traditional cash register and accounting 
machine products. (See, e.g., DX 361, p. 1; DX 370, p. 16.) In its 
1966 Annual Report NCR reported that: 

"The Company's R&D program is designed to achieve two 
basic objectives: 

"1. To improve NCR's traditional position of leadership in the 
control register, accounting machine and adding machine 
markets ; 



^ it "2. To gain for the Company an increasing share of the rapidly 

:| growing market for computer systems and related equipment." 

19 ; (DX 370, p. 16.) 

2$ [j Outside observers also reported on NCR's desire to proceed 

.•I 
2- || gradually in computers. Withington described NCR as following a plan 

22 !i during the 1960s to proceed "methodically" in computers by using 

^ I computers "to complement its existing product and marketing positions". 

^ NCR did this because 
25 ! 

1 

;l -658- 



1 

2 

3 

4 

5 
6 
7 
8 

9 
10 

11 
12 
13 
14 
15 
16 
17 
IS 
19 
20 



"[t]he risks and investments involved in introducing highly 
innovative products to rapidly achieve a major share of computer 
shipments do not appeal to NCR, and that as long as the company's 
overall position, growth, and profit objectives are supported the 
company's computer market share is not a primary objective." (PX 
4834, p. 34.) 

Similarly, an analysis of NCR. by IBM employees in IBM's Market 

Evaluation Department observed that in 1967 NCR was "still in the 

process of establishing itself in computers [and its] management is 

not prone to risk ventures". (PX 2050, p. 4.) 

Developments at NCR during the period from 1964 to 1970 show 

a company improving its products gradually and trying to avoid taking 

the risks of producing innovative products. From 1964 to 1968 NCR 

introduced only a few improvements to its second generation equipment. 

During the summer of 1964, NCR announced a follow-on member of the 315 

family, the 315 Rod Memory Computer (RMC) . (DX 361, p. 14; DX 401, p. 

2.) The 315 RMC used thin-film memory technology and was compatible 

with other computers in the 315 line. (Hangen, Tr. 6 314; DX 361, p. 

14.) Multiprogramming for the 315 RMC was announced during 1966. (DX 

370, p. 16.) During 1965 NCR announced the Series 500 computer, a 

general purpose computer which attempted to combine "magnetic ledger 

bookkeeping with various combinations of punched card, punched paper 

tape or optical equipment". (Hangen, Tr. 10 402; DX 361, p. 13.) 

21 li 

;! NCR also continued to make changes to its existing products . 

|j Despite the fact that the CRAM file had' been superseded by the disk 

ij drive (discussed above at p. 23 5; see Withington, Tr. 56469-70, 



24 
25 



■ ! 



56511) , in 1966 NCR announced a more powerful version of that product 
(DX 370, p. 16) rather than replace it entirely with disk drives. 



-659- 



1 i 

i 

2! 

i 
1 

3 i 

1 

1 
4-1 

1 

5 ! 

1 

! 
i 

1 
3 i 

9 i 

i 

10 i 

i 

12 J 

13 ! 

14 ! 
i 

15 ! 



Following the trend started by System/360, NCR reported in 1966 

that it was increasing the modularity offered by its 315 computer 

family: 

"When NCR announced the 315 computer family, four basic 
processors, three memory sizes, and 12 peripherals were offered. 
Today, 315 users have a choice of nine different processors, 
eight memory combinations and some 60 peripheral units. Expan- 
sion of this flexible computer series in 1965 included three new 
magnetic tape units , three new high-speed printers and a new 
communications controller. This latter device permits up to 100 
communication lines serving input or outDut devices to be linked 
directly to a 315 ► It greatly" extends the power of the system to 
receive inquiries from remote locations and transmit answers . " 
(DX 368, p. 6.) 

It was not until March 5, 1968, almost four years after 

IBM's announcement of System/360, that NCR introduced its third 

generation computers , the Century Series . The first models announced 

were the Century 100 and 200, and NCR stated that it intended soon to 

announce a Century 400 which would be capable of performing time 

sharing. (DX 348, p. 1.) 



,c || The Century 200 was "designed for batch, real-time, and 



17 



scientific processing". (DX 469, p. 2.) NCR offered the Century 100 



1fl II and 200 Systems, in addition to sale, on one, three or five year 
!j rental terms.* (DX 348, pp. 1-2.) Each system was marketed with 

jj 

'\ minimum amount of main memory, a card reader or paper tape reader, 



2o '.! 



-. jj printer and disk drive. (DX 348, p. 1.) Other available peripherals 

. Ji 

;! included CRAM, a MICR sorter-reader, an optical journal reader, 
ii 
-- Ii punched card units, and visual display units. (DX 421, p. 17; DX 



24 -j 

1 * By April 30, 1969, over half of the orders for the Century Series 
" |j were for a five-year term. (DX 372, p. 4.) 

I 

;| -660- 



1 ; 

i 

i 

2 ! 
3 

3 

6 I 
7 

3 

9- 

10 

II ! 

12 

13 

14- 

15 

la 

17 



469, p. 14.) 

NCR promoted the Century Series as its "most important 
new line of products" (DX 366 , p. A), asserting that it incorpo- 
rated many advances over its previous machines, including: 

(1) The Century Series continued the use of thin- film 
main memory introduced on the 315 RMC. NCR called this an 
"important 'first 1 ", making the performance of the thin-film 
memory available at a lower cost* (DX 366, p. 5.) Within 
about a year after the announcement of the Century Series, 
NCR replaced the thin-film memory with core memory. (Hangen, 
Tr. 6329-30.) 

(2) The Century Series used integrated circuits "throughout 
all Century computers and peripherals". (DX 366, p. 6.) 

(3) The Century Series provided for "complete upward 
compatability" so that "as a user's needs increase, more 
powerful processors can replace original units as required". 
(Id., p. 8.) ' 



|t (4) The Century Series included more advanced peri- 

j) pherals — including, for the first time, disk drives: "The 

|| philosophy" of the Century Series "is that the disc concept 

;| is an integral part of all members of the family". The 

jj Series also included a new high-speed printer and, yet again, 



an improved CRAM unit. ( Id. , pp. 6-7.) 

(5) The Century Series had the capability to use both 



23 
24 ; 

-661- 



I! 

! 
3! 

i 

♦■I 

I 

I 

s j 
7j 

i 

3! 
»l 

io! 

12 

13 
14 

15 
16 

17 
IS 
19 it 
20 



COBOL and FORTRAN* programs among others.** 

(6) The Century Series provided for standardization in 

design including standard cabinet frames and panels, power 

supplies and cable connections. It also provided for standard 

interfaces so that "the many peripheral units available with 

Century processors can 'interface' simply, and in a wide 

variety of configurations" . ( Id . ,. p. 7.) 

Of course, while these features represented improvements over 

NCR's prior products, all these features, with the exception of the 

soon to be discontinued thin-film memory, had been included in IBM's 

System/360 four years earlier./ 

In its Annual Report for 1968, NCR announced its marketing 

plans for the Century Series: 

"Over the years NCR has established itself as a 
leading supplier of business systems to thousands of 
manufacturing concerns, construction companies, whole- 
salers, schools, hospitals, utilities, hotels and 
motels, business service firms, and local, state and 
federal government offices. 



* According to Hangen, FORTRAN would be used by NCR users to perform 
scientific applications. (Tr. 10604.) 

** On software development, NCR reported: 



!l 



"Basic computer operating software as well as standard 
Zi |! application programs have been prepared concurrently with 
equipment development. This has insured full program com- 
patibility, plus a proper balance between 'hardware' and 
'software' capabilities." (DX 366, p. 10.) 



22 

23 
24 
25 



/ During this period, perhaps because of NCR's late response to 
System/360, NCR "as a general rule . . . attempt [ed] to price [its] 
products slightly less than the comparable IBM system", that is, "5 
to 10 percent less". (Hangen, Tr. 6350-51.) 

-662- 



1 1 

j 

i 

2 > 

3 ! 

I 

5 I 

6 

7 

3 

9 
10 
11 
12 
13 
14 
15 
IS 



"The advent of the Century Series computer family- 
has multiplied the company ' s opportunities in these 
fields . As users of NCR accounting machines grow and 
their data processing requirements increase, a Century 
100 computer system can meet these greater needs just 
as the Century 200 can serve the larger organization. 
At the same time however, with thousands of new small 
businesses being established each year, the market for 
accounting machines has continued to grow." 



"The largest single market for computer systems is 
in manufacturing^ One out of every four Century Series 
computers currently on order, for example, is scheduled 
for use in this area." (DX 340A, p. 8.) 

Hangen, then Vice President of Finance (Tr. 6233) , emphasized 

in April 1969 the opportunity Century afforded to broaden NCR's market 

ing thrust: 

"Although we intend to continue our close relation- 
ships with the retailing and financial industries, the 
Century allows us to broaden our marketing thrust. We 
are offering specialized Century programs for the 
Educational, Hospital, Local Government (including Law 
Enforcement) , and Distribution Industries." (DX 372, 
pp. 3-4.) 



He added that the majority of orders received for the Century were fron 
17 j| "non-banking, non-retail industries". (DX 372, p. 4.) 

IS Ij The Century Series was "largely responsible" for the fact j 

!j I 

19 ii that in 1968 NCR's domestic orders for computers increased 98% over the 

l| 1 

20 ij prior years. The result of the increase was that "for the first time 

21 ij domestic orders for computer equipment exceeded those for either cash 

22 Ii registers or accounting machines." (DX. 340A, p. 2.) To meet the 

23 '• demand NCR expanded the Electronics Division plant facilities by 50% 

24 'j 

s! 

ij -663- 



and planned a further increase in 1969. (DX 340A, ?. 2.)* 

The large increase in orders had an "adverse impact" on NCR's 

earnings during 1968 as the 1968 Annual Report explained: 

► j- "Users of computer ' systems generally prefer to 

rent rather, than buy such equipment. Thus, the intro- 
duction of a major new computer family such as the 
Century Series tends to have an adverse impact on 
earnings initially, since the company must immediately 
bear production startup, software, training and depre- 
ciation expenses although revenue from rental instal- 
lations is received only over a period of years . " (DX 
340A, pp. 2-3.) 



NCR went on to assure its- stockholders that in -future years 

there would be "a highly favorable effect on earnings". (DX 

340A, pp. 2-3.) Ninety percent of the Century Series systems 

marketed were in fact leased. (Hangen, Tr. 6 358-59.) 

Throughout the 1960s, NCR understood the importance of 

support services — customer training, maintenance, systems design — in 

marketing computer products. It stated in its 1964 Annual Report that: 

"The user of an NCR business system buys con- 
,! siderably more than the machine units which make up 

that system.' In every case, an NCR systems specialist 
and in many instances teams of specialists design 
the most efficient system possible to meet the 
customer's current and future needs, then thoroughly 
train the user's staff in its use. After the system 
3 i is operational, further counseling and assistance 

; l including dependable maintenance are provided." 
G ] (DX 361, p. 8.) 



;i 



"7 



Similarly, it stated in its 1966 Report that: 



5 i * The Century 100 was first shipped in the fall of 1968 (DX 340A, 

| p. B) , and the Century 200 was first shipped in June or July 1969. 
:4 i (Hangen, Tr. 6328.) 



■| -664- 

i 

:i 

'i 

; ! 

! 



1 

2 

3 
4 
5 
6 
7 
S 
9- 

ia 

li 
12 i 

13 
14 
15 
16 
17 
IS 



"Today, a . . . requirement for future success in 
the marketplace has arisen; that is the need for 
business equipment suppliers to provide additional 
guidance to customers in utilization of new technologies 
for operating their businesses more profitably. For in 
the final analysis, the effectiveness of today's 
sophisticated information systems depends upon a full 
understanding of their potential at all levels of manage- 
ment* To this end NCR's educational programs are being 
designed not only to prepare sales representatives to 
install advanced systems, but also to provide counsel 
and training in management sciences." (DX 370, p. 5.) 

And, with the introduction of its Century System it realized that 

customers needed even more support. In its 1969 Annual Report, it 

stated: 

"Marketing requirements of the business equipment 
industry have changed significantly in recent years . 
In recognition of this , the company has taken various 
steps to provide the greater degree of support which 
customers need and expect." (DX 367, p. 4.) 



After IBM announced its "unbundling" in June 1969, NCR's 
Pricing Committee decided whether to make any changes in its pricing of 
support services. (Hangen, Tr. 6364.) Recognizing that there would be 
problems, "particularly [in] customer relations", the Pricing Committee 
did not take any immediate action. (Hangen, Tr. 6393.) On October 1, 
It 1969, NCR announced a change in its pricing structure. (DX 346.) The 



^ ij announcement stated in part 



20 :[ 



"NCR believes that each user of its computer systems 
must be provided with a certain essential amount of 
software, systems support, and educational services if 
)! he is to successfully install the system and begin 
22 !} to benefit from his investment. NCR believes that 

this k as i c package of supporting services must be the 
responsibility of the equipment manufacturer. 



21 



23 ! 



24 



"In addition, NCR recognizes that there is considerable 
variance in the level of support required by different 
customers. This is a function of the capabilities of 



-665- 



f .; 

■ 

i ! 
i 

i 

3 

7 
S 



the customer's Internal EDP staff and of the scope 
and complexity of the applications to be installed. 

"Accordingly, it will continue to be NCR's policy to 
provide,, as part of the basic hardware price, that amount 
of software and support which will realistically insure 
that a prudent user will be able to install and success- 
fully utilize his NCR computer system. 



"Software and support services requested above the 
level which is included in the basic hardware price 
will be separately priced." (DX 346, p. 1, emphasis 
in original.) 

NCR did not change basic hardware prices when it started to charge 

separately for those support services. (Hangen, Tr. 6365.) 

On January 1, 1970, however, NCR partially reversed 

its unbundling decision and announced: 

"After further evaluation, it has been decided not 
to price all basic and applied software and not to 
establish an allowance against which such chargeable 
software would be applied. The NCR software pricing 
plan will be to continue to establish pricing for 
software products on a selective basis , considering 
the value to the customer, uniqueness, and other 
factors. This approach creates an allowance effect 
since the more basic software offerings will not be 
priced." (DX 386, p. 2, emphasis in original.) 



NCR's computer data center business, begun in 1960, expanded 

a :i 

2 i during the 1960s, so that by 1968 there were 69 centers worldwide. 

n \ 

u i Many customers of NCR's data centers used NCR cash registers, account- 

I j ing machines or adding machines to produce "punched paper tape or 

I 
2. i machine readable 'optical' figures as a by-product of normal opera- 

13 : tion." The customers then sent the output media to NCR's data center 

i 
t 

IX I for processing. (DX 340A, p. 10.) The data centers were "NCR's most 

t 
.i 

i5 j successful effort in the data processing business" in the 1960s, 

i 

t 

i 

J -666- 



1 

2 

3 

5 

6 i 
7 

a ! 

i 

9 I 

„i 

12} 

13 I 
I 

i 
I 

15 ! 
16 



according to Withington (PX 4832, p. 22) , and, in addition, proved to 

be a "powerful stimulus to the sale and rental of data capturing" 

devices. (DX 340A, p. 3.) 

NCR's use of its traditional products as input devices for 

its data center computers was an example of NCR's attempt to integrate 

its traditional products with its computer systems. In 1963 NCR 

reported that those products could be used with computers in several 

ways: 

"Many different types of cash registers , accounting 
machines , adding machines and other peripheral units 
are available as basic input devices for [computer] 
systems . Some of these machines communicate with 
computers by means of punched tape or punched cards . 
Others record transactions or other data in slightly 
stylized print which can be read by optical or magnetic 
scanning machines * Still others can be cabled directly 
'on-line' to NCR electronic data processing systems." 
(DX 344, p. 3.) 

NCR did very little, however, in terms of developing and 

marketing on-line systems during the 1960s. During May 1969 H. M. 

'Keller, NCR's Manager of Terminal Communications Products, wrote that 



! in terminal and communication products NCR did "not have a great choice 

ii 

IS ; ' i 

i to offer our prospects", and he listed only one on-line device, the 42- 

11 500, a bank tellers' console. (DX 719, p. i, see Oelman, Tr. 6164.) 

20 : : . 

;! Keller noted, however, that a change had recently occurred in NCR's 

21 ! i 

! ; commitment to on-line devices: 



*^ II "Before we knew that our Company committed itself 

-« jl to creating and offering terminal devices for many, 

lj many purposes , we may have had reasons for not encourag- 
j( ing sales of on-line systems. Now that we know that 
2 ' ;j NCR is committed, each of us must help to penetrate the 
on-line field." (DX 719, p. 1.) 



25 



-667- 



L j In support of that commitment NCR was investing "tremendous 

i 

2,1 sums of money in developing" terminal and communications devices. 

3 | Keller predicted that: 

} ' 

4. ! "To quite some extent, our future success in the 

terminal field will depend upon our success with computer 

5 j. sales and installations » On the other hand, the avail- 
ability of a complete range of terminals will certainly 

gi further enhance our CENTURY sales." (DX 719, p. 1.) 

7 I NCR made a similar prediction concerning terminals in its 

1 
a I 1969 Annual Report: 



10 

11 



n j "More and more people will be brought into direct 

communications with computers through a variety of data 
terminals and data display devices. In fact, it is 
anticipated that by 1975 users of data processing 
systems will be investing as much or more in data 
terminals and related communications equipment as in 

--j the central computer itself. This will create major 
! new opportunities for the business equipment industry 

1- I and particularly for companies such as NCR which has 
j extensive experience in data entry devices." (DX 367, 

l4 J PP. 9-10.) 



2* i And: 



IS ; 

i 

17 J 

13 I 

19 I 

20 : 

21 I 
22 

24 1 
25 



"A decade ago, almost all business machines were 
sold as free-standing equipment. Today, many of these 
products as well as entirely new types of equipment are 
linked together as "total" systems to meet individual 
customer needs. Such systems often include arrays of 
compatible computer equipment including communications 
networks." (DX 367, p. 19.) 



Those predictions turned out to be accurate. During the 

1970s, NCR found that "the capabilities and price/performance of its 

jl terminals [were] an important factor in convincing users to take NCR 

computer systems." (Oelman, Tr. 6183; see also below pp. 998-99.) 

By 1970 it was plain that NCR had oroceeded "methodically" in; 

j 
the computer business, avoiding risks but avoiding also the great j 



-668- 



L 
2 \ 

3 

4. 
f 

r 

3 

9- 

10 

Li 
12 

13 
14- : 



16 | 

17 I 

IS 
19 



success that comes with successful risk-taking. Between 1964 and 1970 , 
NCR's most significant development was the introduction of two models 
(the Century 100 and 200) of a system the principal features of which 
had been available on IBM's System/360 delivered three years 
earlier. With this gradual development, however, NCR reduced the 
chances that it would be a failure like GE and RCA and found itself 
positioned to turn the corner in the 1970s, which it ultimately did. 
NCR's domestic EDP revenues for the year 1969 were $179,298,000, over 
five times its U.S. EDP revenues in 1963. (DX 8224, p. 3.) Even with 
that growth, NCR's domestic EDP revenues accounted for only 26 percent 
of NCR's total domestic revenues but double what it had in 1963. 
(DX 367, p. 6.) 



21 ] 

•i 

22 > 

»1 

2± I 

i 

?c 1 



-669- 



46. Control Data Corporation (CDC) . The period from 1963 to 

1969 was one of rapid expansion for CDC. It added to its two principal 

product lines, the 3000 and 6000 series; it expanded the applications 

. capabilities of its computers to include not only a scientific emphasis 

e j[- ^ut also business -oriented software; for the first time, it developed, 



3 

T 
3 

9 

a 

l 

A \ 

e 

.5 

L7 
LS 



manufactured and marketed a broad line of peripheral equipment, includ- 
ing OEM sales to other companies and IBM plug-compatible equipment; it 
expanded its overseas operations; it made a large number of acquisi- 
tions, including, most importantly that of Commercial Credit Corpora- 
tion, a large financial services company; and, finally, it greatly 
expanded its data center business* CDC's total EDP revenues grew from 
S85 million in 1963 to $570 million in 1969. (DX 298.) Its U.S. EDP 
revenues grew from $83 million in 1964 to $458 million in 1969. (DX 
8224, p. 5.) Its assets increased from $71 million in 1963 to $761 
million in 1969. (DX 302.) To finance that expansion, at least in 
part, CDC raised over $767 million between 1963 and 1969 through equity 
and long-term debt financings. (DX 300.) 

The discussion below focuses on five principal areas of CDC's 
![ growth during the 1963-1969 period: First, CDC's major product 



; announcements, including, principally, its 6000 and 3000 series, and 

20"! 

| their success; second, the increasing use of CDC computers to perform 

21 i 

j both scientific and commercial applications; third, CDC's expanding 

t peripheral equipment offerings , both as attachments to CDC systems and 

22 • 

|as OEM products sold to other companies; fourth, the expansion of CDC's 

7L i 

j data centers; and, fifth, the way in which CDC's growth took place, 

2SJ 

j 

■i -670- 



3. ' 
4 



5 ! 

i 

■ 

7j 

i 
! 

3-! 

»i 

H 
11 1 

12 I 

i 

13] 

I 

14! 

I 
t 

15 ! 

i 

le I 

17 J 
IS I 
19 ! 

zz - 

21 j 

! 
22j 

23 | 

24! 

2S ; 



i.e. , by vertical integration and acquisition. 

a - CDC's 6000 and 3000 Series Offerings (196 3-1969) . 

(i) The 6000 Series. CDC's most important product in the 
1960s was undoubtedly the 6600 computer, announced in July 1962 in 



5 ► connection with a contract let by the Atomic Energy Commission's 



Lawrence Livermore Laboratory, and first delivered in September 1964 
(seven months later than the date contracted for) . (JX 10, 1f 4.) CDC 
Chairman and Chief Executive Officer William Norris* described the 6600 
as a "very great risk" since "it was a trip into the unknown" and 



* William Norris was one of the four founders of Control Data in 
July 1957. (Tr. 5604.) In August 1957 he was elected to the Board of 
Directors and to the position of President of CDC. In 1958-59 he 
assumed the additional title of Chairman of the Board of CDC. At the 
time of his testimony in 1975, Mr. Norris had been Chairman and Chief 
Executive Officer of CDC for over seventeen years. (Tr. 5596-97; PX 
355, pp. 11-29.) Prior to the formation of CDC, Mr. Norris was General 
Manager of the Univac Division of Sperry Rand. (Tr. 5603.) 

R. D. Schmidt joined CDC in 1962 as a salesman. At the time of 
his testimony, Mr. Schmidt was a member of the Board of Directors of 
CDC, Executive Vice President of the corporation and Chairman of its 
Export Strategy Committee. (Tr. 27199-201.) 



. W. Lacey had been employed by CDC for approximately fifteen 
at the time of his testimony, and held the position of Senior 



J 
years at 

Vice President of Corporate Plans and Development. In addition, he was 
Chairman of CDC's Operations Committee and a member of the Board of 
Directors of CDC's Commercial Credit Corporation subsidiary. (Tr. 
6552-53.) 

Gordon Brown, at the time of his testimony, was Senior Vice Presi- 
dent of Marketing and Planning for CDC's Peripheral Products Company. 
(Tr. 50977-78.) 

H. W. Forrest testified by deposition. (DX 13526.) Forrest 
worked in the Univac Division of Sperry Rand under Norris and moved 
with Norris to CDC. ( Id. , pp. 42-44.) At the time he was deposed, 
Forrest was Senior Vice President, Government Relations, for CDC. 
( Id. , p. 4.) 

-671- 



1 i 

i 

2; 

s\ 
4*i 



J I Series computers . ( See Norr is , Tr . 5849-51.) 



5 
7 
8 
9- 

1(1 I 

111 

12 
13 

14,1 
i 

15 ! 

16 ( 

it j 

13 I 

19 \ 

20 ; 



testified that CDC was "betting the future of the company" on it. 
(Norris, Tr.. 5616.) But, as did IBM with System/360, CDC received 
considerable returns on its "bet". Despite early problems with the 
6600, CDC ultimately was successful with it and with the other 6000 



On December 15, 1964, some eight months after IBM's System/ 
360 announcement, CDC formally announced the "6000 Series", then con- 
sisting of the compatible 6400, 6600 and 6800 computers.* (DX 319.) In 
the announcement press release, Norr is described the 6000 Series as 
"the industry's most extensive product line of super-scale compu- 
ters . . . provide ing] business, industry, science and government users 
the most comprehensive range of software and system compatibility ever 
announced in the computer industry." (DX 319, p. 1.) Purchase prices 
for typical 6000 Series systems were announced as ranging "from less 
than $1 million to several million" with rental prices from $25,0 00 to 
$150,000 or more per month.** (DX 319, p. 3.) 

By the end of 1964, CDC had received "possibly five or six" - 
orders for the 6600 (Norris, Tr. 5624), although top officials at IBM 
had believed as early as the Fall of 1963 that as many as 10 accounts 
were then planning to order CDC 6600s. (JX 10, II 9.) Deliveries of 
21 ijthe 6600 were delayed, however, due to unanticipated technological 

li 
22 :i 



^5 ; 



* The 6800, however, was never delivered. (Norris, Tr. 5967; JX 10, 
If 34.) 



** :j *'* CDC later announced two more models of the 6000 Series, the 6500, 
5 - ; i announced in March 1967, and the 6700, announced in May 1969. (PX 355, 

jpp. 36-37; see Norris, Tr. 5626.) Norris described the 6500 as "actu- 
ally two 6400 ! s connected together" and the 6700 as "somewhat more 

I powerful", being "basically two 6600s." (Tr. 5626.) 

-672- 



problems in 1964, 1965 and 1966. (See JX 10, 1! 34; Norris , Tr. 5853- 

54.) By 1966 CDC had solved its technological problems and reported to 

its stockholders on the problems which had occurred in its development 

of large systems: 

" . . * The development and manufacture of very large 
computers are extremely difficult, and severe technical problems 
are inherent in the process. In a past stockholder report, 
it was emphasized that estimating completion dates of very 
large computer developments is becoming increasingly difficult. 
Last year at this time we believed we had found solutions to 
the major technical problems in connection with the 6600. 
Experience since then has proven that, while this was generally 
true, the estimate was in error on the time and effort required 
to make the necessary changes in the equipment and programs. 
The process took longer than anticipated; as a result,, we 
incurred increased penalties for late delivery and retrofit 
costs." (DX 13839, p. 2.) 

CDC also found it difficult to establish a price for the 

6600. In April 1964, CDC submitted 6600 proposals to the Bettis Atomic 

Power Laboratory (BAPL) and the Knolls Atomic Power Laboratory (KAPL) , 

in competition with IBM, Burroughs, Philco and Sperry Rand proposals.* 

(JX 10, U 12; Norris, Tr. 5620; DX 4960, p. 5.) The bidding process 

^ i! 

1 was highly competitive. Initially, BAPL and KAPL selected CDC and 

IT!! 

; Burroughs resDectively . Later, however, both BAPL and KAPL changed 

IS 

; their selections to IBM. (JX 10, fl 12.) 

IS 

;[ Six months later, CDC was told by the Government that it was 

20 

22 

22 

24 

25 ' 

! * IBM proposed 360/90 's, with the interim installation of Model 70s 

] until the 90s were ready for delivery. (JX 10, if 12.) 

••{ 

.{ 

I -673- 



L 
Z 

A- 

4- ft 
7 

a 

Iff 

n 

12 

13 
14- 



interested in reopening the BAPL and KAPL negotiations if CDC was 
prepared "to sharpen [its] pencils". (Norris, Tr. 5620.) According to 
CDC Chairman Norris, BAPL and KAPL then misled CDC "in a deliberate 
manner" as to the terms of the I3M offering, telling CDC "that IBM had 



L I offered a computer at four times the power of the 6600 at a lower 
Z\ price", as well as misrepresenting the date at which IBM could deliver 
3 j its equipment. (Morris, Tr. 5970-73; PX 367, p. 5.) CDC made an 
4- 1 "unsolicited proposal" to BAPL and KAPL in late February 1965 "at a 
5- |. price substantially lower than that previously proposed by CDC and 
5; substantially lower than the price proposed by IBM."* (JX 10, K 12; 
7j DX 324.) CDC "proposed a combination deal which would involve replac- 



ing the 6600 within some period of time . . . with the computer that 
would be much more powerful than the 6600, . . . the 6800, and at the 
time the 6800 was delivered, that we [CDC] would take back in trade 
the 6600." (Norris, Tr. 5621.)** 

CDC, "unfortunately" according to Norris, ultimately won the 
BAPL and KAPL contracts. (Tr. 5963, 5976; JX 10, If 12.) Moreover, it 
was unable to meet the delivery dates and, as a result, was required to 



* CDC had earlier reduced the price on the 6600 because of "substan- 
tial reductions in prices of component parts (transistors, diodes, 
ijetc.) which . . .'occurred in [1963 and 1964]". (DX 13838, p. 4; 
Schmidt, Tr. 27416.) 



3 

9- 
10 
11 
IZ 
13 
14- 
15 
16 

17 ! 

13 •. 

ii ** One other aspect of CDC's pricing policies is worthy of note. Botifi 

.q 'I Norris and Lacey testified that, as a general rule, CDC set the prices 
it for its computer systems five to ten percent below IBM's prices. 

2_ II (Norris, Tr. 5653; Lacey, Tr. 6567-70.) According to Lacey, this was 
ji "[b]ecause our [CDC's] experience tells us that if we on a grand aver- 

-- ;| age basis go significantly higher than that, that our opportunity for 

^ j business rapidly diminishes". (Tr. 6573.) CDC considers the "prices 

I of all manufacturers but principally the prices of IBM and, secondarily, 
22 ij of other manufacturers". (Lacey, Tr. 6569.) Norris and Lacey admit- 
-3 ij ted, however, that it is extremely difficult to compare accurately the 
^ 1 performance of the system of one manufacturer as against that of another 

; | manufacturer. (Norris, Tr. 6038-40; Lacey, Tr. 6800-01.) Norris 

24 ij testified that computer companies compete on the basis of a variety of 
!{ factors other than price; reliability is a factor, for example. 

25 ii (Norris, Tr. 6040-41.) 

; i 
'! 

II -674- 



i 

z\ 

i 

31 

i 

?! 

7 j 

3 

a 

10 
U 
12 



pay substantial penalties which further reduced its effective price. 

The final settlement was "substantially disadvantageous to Control 

Data". (Norris, Tr» 5976-78.) For a time, difficulties with the 6600 

\ adversely affected CDG. According to Norris: 

"We were losing money as a company in 1966/1967 primarily 
because of problems with the 6600 computer. Frankly, there 
was a great deal of conflict in top management in 1966 over 
whether we should press forward or retrench — closing down 
data centers was high on the list of retrenchment possi- 
bilities. The decision was made to press on, however 
there were some deserters in top management as a result — 
they were afraid that the ship was sinking." (DX 284, p. 6; 
see also Norris, Tr. 5678-79.) 



Mr. Norris also testified that CDC "had to rush into the 
6600" because it had been "literally clobbered by IBM competition" to 
CDC's earlier 1604 computer system, and that with the 6600, CDC again 
faced the "enormous impact of competition from IBM". (Tr. 5625.)* 

f Ultimately, however, the 6600 — and the 6000 Series in general — proved 
to be "particularly" successful for CDC. (See Norris, Tr. 5849-51.)** 
In 1969, for example, CDC successfully bid for an Air Force procurement 

ij to replace Univac- and IBM equipment with 12 or 13 6000 Series machines,. 

17 j 

i * Such militant language by persons speaking for firms in a competi- 
IS ;j tive environment is not uncommon. For example, CDC's chief development 

:j engineer for the 6000 Series, Vice President Seymour Cray, at CDC's 
2Q j June 1963 corporate planning meeting, urged that CDC announce the 6600 

J and a successor in order to "slug" IBM because, in his opinion, IBM had 

21 ■! "made a mistake by putting all [its] eggs in an integrated circuit 
!| basket". (DX 13526, Forrest, pp. 748-750.) 

22 1 

j ** CDC received more than $286 million 'in revenue and more than $185 

23 j million in gross profits from the 6600 computer systems during the 
[period 1964-1972. CDC's gross profits on the 6600 exceeded its gross 

2*. 'j profit objective which was set to yield a reasonable rate of return on 
j investment and a reasonable net profit at the bottom line. (DX 1185, 

?s i pp. 3-5.) We are aware that DX 1185 is an offer of proof and not 
1 evidence; however, we rely on this offer of proof because it is 
;l consistent with the other evidence about the success of the 6600 
:j systems and the growth of CDC. 

I -675- 



Li including several 6600s. The contract was for systems "to handle the 
Z '■ entire inventory scheduling acquisition of spare parts for the Air 
j jj Force Logistics Committee" and its aggregate value to CDC was approxi- 
4. if mately $40 million. (Schmidt, Tr. 27469-76.) 

CDC finally manufactured 94 6600/6700* computers (as compared 
to some 17 360/90s manufactured by IBM, including four for use within 
IBM) and a total of 215 CDC 6000 Series computers. (JX 10, 1f1f 35, 36.) 
In the late 1960s, as a successor to the 6600 and a replace*- 
ment for the never-delivered 6800, CDC developed the 7600 computer, 
which it officially announced in December 1968 and first delivered the 
following month — more than 21 months after the first committed delivery 
date for a 6800 and seven months later than the delivery date called 
for in the first contract using the machine designation "7600". 
(Norris, Tr. 5628; JX 10, 1f 34; PX 355, p. 39.) Norris characterized 
the 7600 as "several times more powerful than the 6600 and it addresses 
the same market". (Tr. 5628.) CDC installed its first two 7600 compu- 
ter systems during 1969 (DX 13843, p. 4) and in that same year, CDC 
Vice President J. W. Lacey, speaking to a CDC graduate orientation 
class, described CDC's success as follows: 



3: 

a 
7 ; 

a 

9 

a 
x 
,z 

3 
L* 

^ , 
17 ;; 
IS • 

19 ii 

;| " [W]e have a world-wide leading position in large computers 

2C i today. That position is widely recognized. Since 1964, with 

j the delivery of the first 6600 Computer, followed recently by 

21 j the 7600 Computer, Control Data has dominated this market, 
j Second, there is a rapidly increasing trend towards very 

22 ;| large computers used in data processing networks in which 
j many users share the enormous power of machines like the 

22 ; 6600, and away from medium sized and small sized stand- 
alone computers. ..." (DX 43 8, p. 7.) 

24. } 

i 

25 | 

! * The 6700, announced in May 1969, was "basically two 6600's". 

j (PX 355, p. 37; Norris, Tr. 5626.) 

j 

; -676- 

i 
•i 

•i 
i 
t 



L 
2 

Z 

c 
7 

a 

10 

n 
iz 

13. 
14- 



Looking to the future of the company, Lacey said " [w] e believe that 
our position today and the direction we are giving our business puts us 
in an outstanding posture to share in the explosive future growth of 
our industry 1 ** ( Id. r p.. 13; see also Lacey, Tr. 6676-77.) 

(ii) The 3000 Series > The 6000 Series was not the only 
product line developed by CDC in the 1960s. CDC also significantly 
expanded its 3000 Series. In September 1963/ CDC announced its 3200 
computer; in January 1964, it announced the 3400. (PX 355, p. 34; 
Norris, Tr. 5627.) Norris testified that the 3200 competed with "IBM., 
Univac, Burroughs, NCR to an extent, and possibly SDS". (Tr. 5627.) 

1965 saw the continued expansion of CDC's 3000 Series. 
The 3300 was announced in November 1965 and delivered in that same 
month. (PX 355, p. 35.) Norris described the 3300 as CDC's "entry 
into timesharing. And, again, I think it had some added features for 
business data processing. And it was a considerably lower-priced 
machine than, say, the 6600. It was what you term then a medium- 

l[ size computer." (Tr. 5627-28.) According to Norris, it competed with. 

| "IBM, Univac, Burroughs, SDS and NCR." ( Id. ) 

; The 3500 was announced in November 1965, although it was not 

19 J 

:| delivered until 1969. (PX 355, pp. 3 5-3 6.) Norris testified that it 

2C !| 

; was "essentially the same computer" as the 3300 except for the use of 

21 i| . 

:• integrated circuits and "somewhat larger memory options". (Tr. 5628.) 

22 1 

In 1967, CDC announced its 3150 computer, the smallest of the 

22 i 

i 3000 Series, stating that it "provides a complete business and scienti- 
i 

2- i 

';} fic information handling capability with a minimum of hardware and 



-677- 



software. The. 3150 provides maximum throughput at low initial cost to 
the user and the capability for him to expand upward as his information 
handling needs grow.." (DX 13340, p. 8.) 

The 3000 Series was a substantial success for CDC, without, 
in large part, the start-up problems that beset the 6000 Series. 
Indeed, as early as 1966, CDC was able to describe its 3200 — which was 
introduced little more than two years earlier — as "highly successful" ► 
(DX 13839, p. 5.) And in 1968 CDC reported to its stockholders that 
"orders for our 3000 product line continue to increase-both in the numbejr 
of systems ordered and in average dollar value." (DX 13842, p. 2.) 
The 3000 Series was successfully marketed for applications in manufac- 
turing, general business data processing, education, medicine, data 
services and the brokerage business. (DX 13843, p. 4.) Moreover, it 
gave CDC a lower-priced alternative to the expensive 6000 Series 
computers. 

b. CDC's Expansion into Commercial Data Processing. As the 
:j decade began, CDC perceived itself as offering large, "scientific" 

r ;{ 

; | computers. Very quickly, however, CDC learned that the distinction 

H 
: between scientific and commercial data processing — if indeed there ever 

■j was on— had blurred almost to the vanishing point, and by the end of 

* '< \ 

| the 1960s, CDC estimated that fullv 40 percent of its business came t 

N - ! 

• from "pure business data processing". (Schmidt, Tr. 27476-78.) | 

♦ CDC's Chairman Norris described CDC in the early and mid- ; 

" 1 i 

! 1960s as "a supplier of large-scale digital computers to scientific and ! 
i engineering applications". (Tr. 5624.) Norris also testified that, at 



-678- 



L 
2 

Z 

4* 

5 

a 

7 

3 

S- 

LO 

II 

12 

13 



IS;! 
17 I 
IS 



the time of its announcement in 1962, CDC thought that the 6600 "would 
be unique to a great extent . • . it being so much more powerful and 
so well-suited to scientific work, it would just be outstanding in the 
eyes of those laboratories that have these very large scientific pro- 
blems". (Tr. 5617-18.-) 

CDC Vice President, Gordon Brown, described the entire 6000 
Series as announced as 

"very definitely a scientific line of computers, and there- 
fore, the analyses that we did showed that the strength of 
the 6000 product line prevailed . . . over IBM and Uni vac in 
most typical environments; and, on the other hand, proved to 
be deficient when it was employed in an environment requiring 
a lot of input/output of data, or commercial type requirement 
[because] the architecture of the 6000 series was designed 
with the scientific user in mind. It had a large, fast, 
central processor with a number of auxiliary processors to 
handle the input/output functions. And it had a large, very 
fast disk storage capability associated with it." (Tr. 50996.) 

By October 1965, however, the CDC Executive Council* had 

recognized that there were no longer separate markets for scientific 

and business data processing. (DX 276; see Norris , Tr. 6002-06; Tr. 

6081-82.) Thus, between 1964 and 1968, according to Brown, "gradually 

additional capabilities were added to the 6000 computer system, and 

these included COBOL compilers of sort and merge packages and the 



ability to handle permanent files as opposed to using the input/output 



2C ; 

•i 



devices as auxiliary storage or temporary storage of data files." (Tr. I 

TT .1 1 



;j 53064-65.) } 

22 j . | 

Similarly, CDC Vice President Schmidt testified that although ; 

22 ; i 



24 \ 

j * Lacey described the CDC Executive Council as "responsible for 

25 j advising our Chief Executive Officer concerning major business ques- 

tions". (Tr. 6556.) 



-679- i 

i 



t 



" [i]n the early stages of the 6000 marketing effort, we aimed at primar-' 
ily the scientific applications . . . that has changed." The change 
"started with the coding of a COBOL compiler for the 6000 Series and 
the 6600 specifically. " (Tr. 27457.) CDC described the introduction 

; of a COBOL compiler for its 6000 Series in 1967 as "an important 
achievement, for we are now able to provide our customers with the full 

I power of our super computers to handle the broad scope of their data 
processing problems." (DX 13841/ p. 4.) Then came the development of 
application programs for the 6000 Series in COBOL and the sale of "some 
limited number of [business] applications, usually in conjunction with 
primary scientific applications." By 1968, CDC had sold 6000 Series 
systems in Mexico "which were devoted primarily to business data pro- 
cessing, using that COBOL compiler and the COBOL application programs." 
And in other situations, customers with business applications as well 
as scientific applications ordered a 6600 system to do both. (Schmidt, 
Tr. 27457-58.) 

The primary impetus for the broadened use of the 6000 Series - 
came from customers who wished to have a single machine capable of 
performing both commercial and scientific applications — one of the 
primary reasons that led IBM to develop with System/360 the capability 



l to do both applications equally well. (Norris, Tr. 5618; see JX 38, 

1 I • 

.1 pp. 27-29; see pp. 290-96 above.) As Norris testified: 

2 ;'» 

j "We found that there were large companies who, while 

3 the majority of the work that they wished to do was of an 
engineering and scientific nature, still they had a certain 

\i, j amount of business data processing and that they preferred 

i to have only one computer as opposed to having two computers , 

?c i one for scientific and the other for business. 

I 
.1 

1 -680- 

i 



L 
2 
Z 

4>j 

5 
& 
7 

a 

si 
ia 

tl i 

i 

12 i 
13 

15 



"So, we set about to broaden out the software which was 
available with the 6600 so that we could meet the requirements 
of those customers where the bulk of the work was still scien- 
tific but still the 6600 would do the business data processing 
well enough so that the customer only had to have the one 
computer. " (Tr. 5618 . ) 



While CDC also introduced the 3000 initially as "basically 
scientific", it realized from the outset that the 3000 Series "had a 
little bit more versatility as a business data processing machine than 
the 6600". (Norris, Tr* 5627.) Over time, CDC added hardware features 
and software packages to enhance the 3000 Series desirability for 
business applications. (Brown, Tr. 50990-91.) 

Among the uses of the 3200 during the first years of that 
system's life were: medical research and training and use in "flight- 
testing ground stations" (DX 13838, p. 7); and use in combination with 
a 3600 "to integrate the computing and business data processing" for 
the 57 associate companies of Phillips, the Dutch manufacturer. ( Id. , 
p. 9.) 



• a ;! In 1965, CDC announced the 3300 which, according to Norris, 

17 |( "had some added features for business data processing." (Tr. 5627; 
IS jj PX 355, p. 35.) In its 1967 Annual Report, CDC stated that "the 
IS it variety of applications being handled by the 3300 include production 

20 ■•' scheduling, labor analysis, data communication, inventory control, 

.1 
ZL | engineering computations, and general business data processing." 

22 1 (DX 13340, p. 8.) And by 1966, according. to Brown, "the 3000 product 

23 i line . . . was evolving to a . . . better balanced product line between 

24 j both the scientific and the commercial users. The initial base of 
.! 

25 I customers had largely been scientific users, and many of them were 

■I 

;i -631- 



1 

2 

Z 



4- [GDC did not manufacture- its t own peripheral equipment. (Norris, Tr. 



J 
6 

7 

i 

i 

s; 

! 

?! 

i 

u j 

12 j 

13 I 



starting to expand their applications for commercial usage." (Tr. 
50-9-97/ see Tr... 50990.) 

c. CDG's Expanding Peripheral Business. In its early years 



5609r see PX 6066, p. 1.) In the 1963-1969 period, however, CDC began 
to manufacture peripheral equipment not only for attachment to its own 
processors but also as an OEM supplier for other EDP companies. In 
addition, CDC laid the foundation for its later very successful entry 
into the IBM plug compatible peripherals business. 

By the early 1960s, CDC had recognized that the sale of 
peripheral equipment was potentially a highly profitable opportunity 
and therefore began to expand its peripheral offerings. For example, 
in 1964, in its news release announcing the 6000 Series, Norris stated: 



t4. | . " [Nlumerous peripheral devices . . . are under development 
and will be announced over the next two years to complete 
the implementation of products required for total manage- 
ment information systems. These peripheral devices include: 



15 
16 
17 
IS 
19 
20 

;l 

-, :! "Optical character recognition readers. 
"Line of visual displays. 



"Disk files — not only low-cost units but very sophisticated, 
high capacity, low access time, extremely high transfer rate, 
mass memories. 

it "Mass core memory. 



"Remote terminals and processors for on-line man/machine 
interaction. 



22 

23* 

24 

25 



"Line printers, card ounches and readers." 

(DX 319, p. 3.) j 

In 1965, CDC acquired Data Display, Inc., a manufacturer of j 

"electronic display peripheral equipment". (PX 355, p. 5; DX 296.) In j 

I 

I 

-682- i 



1965, CDC announced its 852 disk drive which was "in many ways like the 
IBM 1311". It was marketed "in a very modest way" on an OEM basis to 
GE and Honeywell in Europe (Brown, Tr. 51015-17) because "there was 
very little market demand for this type of product at the time, and CDC 
was just beginning to build and staff an effective market and service 
organization." (Brown, Tr. 51015-51017.) 

In 1966, CDC announced its 9433/34 disk drive — an IBM 2311- 
type device, although it was not media compatible with the 2311 drive — 
on an OEM basis, with first shipments occurring in 1967. CDC's princi- 
pal OEM customers for the 9433/34 were Honeywell, GE and RCA, as well 
as ICL in Great Britain and Siemens in Germany.* CDC eventually sold 
some 16,000 94 33/34 drives in the late 1960s on an OEM basis, at prices 
less than one-half that of the IBM 2311. (Brown, Tr. 51056-58.) The 
development and marketing of such IBM-type devices, of course, fore- 
shadowed CDC's later decision to produce IBM plug-compatible peri- 
pherals. (See Brown, Tr. 51063-67.) 

In the 1966-67 time frame CDC also began to market peripherals 

I originally designed for its 6000 Series on an OEM basis, such as the 

it 

; 6638 disk file, which was sold to Honeywell, ICL and GE as the 94 90.** 



22 
23 
24 
25 



"We view the computer market as a world market , and plan our 
organization and operations to maximize our abilities to best 
satisfy our customers in that market." (DX 14214, P- 5; emphasis 
in original.) 

** Less than 40 9490 disk files were shipped, however, according to 
Brown, because there was not a large market at that time for that type 
of fixed non-removable storage disk device and because not many OEM 
customers had channels that could take the high data rate of the 9490. 
(Tr. 51033-34.) 

-683- 



} 

r ! 



(Brown, Tr. 51032-34.) In 1967, CDC introduced a new line printer, new 
tape transports, a card read-punch, a magnetic drum storage unit and 
several new versions of electronic display terminals. (DX 13840, p. 
8.) 

During 1968, CDC added to its peripherals product line "a 5 
billion bit disk file, a 1200 line per minute printer, and a new gene— 
r i| ration of tape transports"- (DX 13841, p. 8.) At the same time, it 
informed its stockholders that " [independent suppliers and the in- 
house developments of major computer manufacturers do and can be 
expected to continue to intensify competition." ( Id . , p. 8.) 

And in 1969, CDC introduced six more new peripheral products: 

a disk storage unit, two printers, a card reader, a display terminal, 

m -\ and a drum device. (DX 13843, p. 6.) Also, CDC announced an IBM 2314- 

.type device in 1969 for use with CDC ' s 3000 Series, 6000 Series and 

CYBER 70 product lines. (Brown, Tr. 51068-69.) 

By the end of the 1960s, CDC "had made major investments in 
5 ;! 

;[ technology in most of the principal peripheral areas. This started 
7i| 

j with the development of subsystems for use in [CDC's] own computers and 

■ carried through most of the Sixties . . . into the development of a 

- j j 

5 fairly large base of OEM business." (Brown, Tr. 51212; DX 4 38, p. 12.) j 

j 

j d. Data Centers. CDC also greatly expanded the data center i 

2. 1 - ! 

\ (service bureau) portion of its business in the 1963-1969 period. In j 

'2. :S i 

~ 11964, for example, CDC had six data centers. (DX 13838, p. 8.) By j 
"* !l969, it had more than 40 "throughout the world". (DX 13343, p. 6.) j 

I The six data centers oaerated bv CDC as of 1964 used CDC 36 00 

.( *■ ■* i 

?e ■ ! 

~* I ! 

I j 

:j -684- I 

;■ i 

"» i 

i I 



L 

2 

3 

4 

5 

S 

7 

3 

& 

10 

II 

12 

13 

14 

15 



and 1604-A computer systems, forming a network — later known as 
CYBERNET — "tied together by Bell System Data-phones" and "providing 
complete data processing services to commercial and government users oi 
a contract basis." (DX 284, p. 4; DX 13338, p. 8.) Typical of the 
many applications processed at the centers were "Operations Research 
applications", "traffic surveying and planning", "Hospital data 
processing" and "school scheduling and grade reporting". (DX 13838, p, 
8.) 

By 1965, CDC had seven data centers and had begun its "net- 
work development". (DX 284, p. 4.) However, "[e]xcept for brief 
periods in the mid-6 0's, data centers in the aggregate operated at a 
loss until 1972 because [CDC] kept pouring money into expansion". 
( Id. ) * Norris, in a draft of a speech in 1973, cited this as an 
example of CDC's "willingness to take risks". ( Id. ) 

In the fiscal year ending June 30, 1968, CDC acquired C-E-I-I 
•g jia company which offered computer programming and other professional 

data processing services, and Pacific Technical Analysts, Inc., claimed 
to be "the largest and most capable programming and service center 
'{company serving the Western Pacific area". (DX 13841, p. 2.) By the 
j end of fiscal 1968, CDC was operating over 30 data centers worldwide 



i 
17 



18 



21 ! 

22 

23 

24 

25 



and offering "an extensive inventory" of application programs. (DX 
13841, p. 7.) 

In 196 9, CDC offered the following description of its 
CYBERNET network of data centers: 



* Accounting losses which result from expenditures made for the pur- 
pose of achieving later returns, are, of course, not truly economic 
losses but rather investments. 

-685- 



I i 

i 

z\ 

I 
5-1 

e 

7 
8 
9- 

L0 

n 

L2i 

13 

14 

15 

16 



"Through the highly advanced CYBERNET service, customers 
have convenient access to the cost /performance advantages 
offered by both the CDC 3300 and 6600 computers without having 
to make large capital outlays." (DX 13843, p. 6.) 

With more than 13,000 miles of communications lines by the end of 

fiscal 1969, CDC was offering its data services at more than 40 data 

centers throughout the world, a more than 550 percent increase over its 

1964 holdings. ( Id. ) 

e. CDC's Acquisitions (1963^1969) . The story of CDC's 
expansion in the 1960s cannot be fully understood without considering 
CDC's acquisitions during that period. Between 1963 and 1969, CDC 
acquired some 43 companies, at a total cost of over $897,000,000. 
(Norris, Tr. 5788-89; PX 355, pp. 3-9; DX 296.) 

All of those companies — with the exception of Cedar Engineering, 
Kerotest and Commercial Credit Corporation (which is discussed in some 
detail below) — were supplying an EDP product or service at the time of 
acquisition. (Norris, Tr. 5794-95.) 

CDC's numerous acquisitions, most of which were paid for by 



17;JCDC stock (Norris, Tr. 5789), enabled it to broaden its product and 

» 

18 ; service offerings quite rapidly without the substantial development 

! 

19 ; time that internal expansion would have required.* As Norris stated in 
2G :;a draft of a 1973 speech: 



n 



T ii 



22 jj * That is not to say that CDC did not expand internally as well, 

I particularly through increased vertical integration. For example, CDC 

23 ij decided in 1966 to, have its research division manufacture integrated 

;j circuits for use in the prototype of the 3500 computer rather than buy 
24. Ii circuits from Texas Instruments. (DX 432.) Also in 1966, CDC reduced 

;• costs by bringing the manufacture of card module assemblies, memory 
25 J cores, memory planes, memory stack assemblies and logic chassis assem- 
blies in-house. (DX 13339, p. 9.) 

'i -686- 



"Our high P/E ratio stock, or Chinese money, as 
we often termed it, was used to acquire companies with 
complementary technology, products, services and markets. 
In other words, we were not trying to broaden our base 
as in a conglomerate, but rather to buy new computer 
products and services and markets to spread development 
costs- and gain economies of scale as rapidly as possible." 
(DX 284, p. 7.) 

Norris agreed that this was "an alternative to investing money in 

research and development" and was "very successful for Control Data for 

that purpose". (Tr. 5804-07.) 

Speaking about the acquisitions in general, Norris testified: 

" [W] e wanted to have our own peripheral equipment 
to put on our computer systems so that we would have full 
control over the cost and quality. We wanted to broaden 
out our product line both with respect to hardware as 
well as software. 

"In some instances we bought data services businesses, 
which gave us additional revenue and profit. And we were 
able to take those services in turn and have them sold by 
a larger marketing organization." (Tr. 6092-93.) 

Thus, in the fiscal year ending June 30, 1964, "Control Data 
made significant additions to its technical capabilities and product 
lines, and broadened its market areas" by way of a number of acquisi- 
tions. It acquired companies with capabilities in the areas of: 
digital computers for use in power, chemical, petroleum and oil indus- 

is i 

:[ tries; card punch and reader systems and other peripheral devices; 

2G ! 

! optical character recognition equipment; data collection systems,- data 

21 : i 

:! processing services; printers; and analog to digital conversion equip- 

22. "> 

■j ment. (DX 13838/ p. 5.) 

22 1 

j In the fiscal year ended June 30, 1965, CDC acquired 

24, :\ 

j companies with capabilities in electronic display devices and pro- 

2 C i 

j gramming consulting services, as well as a business data processing 

i 

i 

1 

I -687- 



i 
2 ; 

i 
i 

e 

7 

a 

9- 

12. 

12 

14- 
15 

16 i| 

17 ;! 

is 



1 
Z 

3 

4- 

5 

t\ 

sj 

9 I 

ia j 
i 
i 

11 : 
13 

IS 



center and two companies whose products, involving radar, for example, 
incorporated the use of digital computers. (PX 355, pp. 5-6.) 

In the year ended June 30, 1966, CDC acquired the commercial 
computer operation of General Precision Inc.'s Librascope Group. 
"Included in the purchase were General Precision's commercial computer 
rental and service contracts, and inventory of commercial computers" as 
well as its "highly experienced commercial computer sales and service 
organization. " Other acquisitions were of an electronic systems 
engineering company, a Hong Kong firm doing assembly of electronic 
components, particularly ferrite cores, and an Italian firm operating 
data centers in Italy. (DX 13839* p. 3.) 

In 1967, CDC acquired, among others, C-E-I-R, (DX 296, p. 2) 
because, as Norris testified: 



" [C-E-I-R] had the American Research Bureau, which 
was using a computer in surveying the listener response 
in the television and radio industries; there was Automation 
Institute — these are schools to teach computer programming 
and computer operation; there was a data services business 
IS ;j .... And it was primarily those three areas that were 
particularly interesting to Control Data." (Tr. 5796-97.) 



17 

IS 
19 j 



CDC Vice President Lacey testified that the acquisition of C-E-I-R "was 

an additional entry for Control Data into the data services and consult- 

i 
I 

ing services business, beginning steps of our broadening or our business! 



J line". (Tr. 6632-33.) I 



j CDC's single most important acquisition — Commercial Credit — 

j occurred in 1968. (PX 355, p. 3.) In August 1968 CDC acquired 

43 i 

! Commercial Credit, "a diversified financial institution . . . 

2*1 

| with nationwide and Canadian operations in financing, lending, leasing, 

25 ! 

'.factoring, and insuring" (DX 13342 , p. 16) for 4,825,720 shares of CDC 



-688- 



stock, with a total market value of $745,573,740 (PX 355, p. 8) — by far 
the most expensive acquisition made in the 1960s by Control Data. 
( Id. t PP» 3-9.) The principal reason CDC acquired Commercial Credit 

■■ was to gain a financial services subsidiary in order to enable CDC 
better "to finance computer leasing." (Norris, Tr. 5643; see Lacey, 
Tr. 6586-88.) 

Initially, CDC marketed its computer systems on a purchase- 
only basis. However, by 1961 or 1962, CDC had realized that many EDP 
customers demanded leases and, accordingly, it began to offer its 
system for lease as well as purchase. (Norris, Tr. 5641-42.) Hence, 
CDC over time has offered one-year, three-year, five-year and longer 
leases. (Norris, Tr. 5644.) CDC first offered three-year leases in 
1966 and non-cancellable five-year leases in 1967, both at a discount 
from its short term lease price. (DX 295.) 

CDC's changeover to leasing as well as purchase required that 

|j CDC "raise additional sources of capital . . . [ijn order to finance 
the leases [because] when you lease a computer you get paid on a 
monthly basis , but you have to incur the total cost of the computer at 
the time it is delivered." (Norris, Tr. 5642-43.) Thus, in 1966, CDC 



L 

Z 

z, 

4> 

a 

6 

7 

3 

ia 

u 

12 

13 
14- 
15 
15 
17 
IS 

19 ;i 

it entered into an arrangement with Leasco whereby Leasco would purchase 

20 : 

| CDC systems and then lease them to customers on a long-term basis. (DX 

21 "\ • 

j 13839, p. 2.) However, CDC cancelled that agreement the following year 

22 "» 

j "in light of current and prospective financing plans of the Company". 

23 j 

(DX 133*0/ p. 12.) According to Norris, it was not until CDC acquired 

2*1 

:j Commercial Credit Corporation in 196 8 that CDC ultimately "solved the 

25 '■! 

[problem of financing leases". (Tr. 5643-44.) 



-689- 



f.. Conclusion. When the 1960s began, CDC was a virtual new- 
entrant in the EDP industry, having been incorporated only three years 
earlier* (DX 271,, p. 7.) It perceived, itself as being principally, if 
: not solely, a supplier of large scientific computers. (Norris , Tr. 
5624,) By the end of the 1960s, CDC was firmly established as a major, 
diversified competitor in the EDP marketplace. It had achieved great 
success with its 6000 Series computers. It had added considerably to 
the business data processing capabilities of its computer offerings , to 
the point that by 1969 approximately 40 percent of its business came 
from "pure business data processing". (Schmidt, Tr. 27477-78.) It 
had greatly expanded its peripheral equipment offerings and begun a 
successful business as an OEM supplier of peripherals. And it had 
increased the number of its Data Centers from six to 40. (DX 13838, p. 
8; DX 13843, p. 6..) The 1960s were indisputably a period of great 
success for Control Data Corporation. 



»! 

■\ 

i 

\ 

":i 



- '! 



-690- 



z 

3. 

5" 
5 
7 
3 
9 

ia 

12 
13 



4-7. SDS .. Scientific Data Systems ("SDS") was formed in 
1961 with an initial capitalization of approximately $1 million, 
raised from a San Francisco venture capital company and the firm's 
original founders. (Palevsky, Tr. 3128, 3193*)* SDS was the idea of 
Max Palevsky who furnished approximately $60,000 to $80,000 of its 
initial capitalization: "I put up half the money in cash and half as 
a note." (Palevsky, Tr. 3127, 3193.) For his investment, Palevsky 
received "something in excess of 15%" of SDS * s equity. (Palevsky, 
Tr. 3193-94.) 

Palevsky had begun his business career as a research analyst 
for the Bendix Corporation in 1952, in "the division of Bendix that 
was starting to explore computers". His responsibility was 
"[plrimarily logic design, computer design .... It was a time when 



T £ :■ everybody did everything". Upon leaving Bendix in 1956, Palevsky 
.- : organized the Packard Bell Computer Corporation, a subsidiary of 
. £ ! Packard Bell, which "buil[t] specialized digital computers, special 
purpose digital computers, and eventually a small general purpose 



u 



iS ;i 



} * The witnesses testifying about SDS/XDS (SDS was called XDS — 

i - ;[ Xerox Data Systems — when it became a division of Xerox) were Max 

| Palevsky (described above) , Harvey Cohen and F. R. Currie. Only 

23 j Palevsky was with SDS in the early years. Cohen arrived about 1964 

i and held a number of positions, including in 1967-68 the Director 

21 | of Marketing Operations. (Cohen, Tr. 14427-28.) Currie also came 

-j in 1964 and held various marketing positions, becoming Vice-President 

22- : ] of Sales in 1968-69. (Currie, Tr. 14909-13.) At Xerox Data Systems, 

•j Cohen became Vice President of Advanced Systems, Business Development 

23. j Group (Cohen, Tr. 14427-28, 14521) and Currie became Vice President 

j of the Data Processing Division reporting to Cohen. (Currie, Tr. 

2- 'I 14917, 14922-23.) Currie later moved to the Corporate Marketing 

! staff. (Currie, Tr. 14923-24.) 



i -691- 

•t 



1 !, computer and digital systems".* (Palevsky, Tr. 3121.) In 1961, 

2 : Palevsky left Packard Bell because 

3 ;| "that company had come on hard times. The ideas I had about how 
J to proceed in the computer industry required much stronger 

4 j| backing from the parent company which they could not provide. 

I also felt that the computer industry is a very unique 



51; kind of industry, and it was very difficult working under a 

|| management that really knew nothing about the industry itself, so 

5 jj that it made sense to be independent, and, of course, there wera- 

jj opportunities to make a great deal of money". (Palevsky, Tr. 

7 jj 3127-28.) 
•i 

8 



9 
10 
11 



SDS initially conducted all its activities in a 5,000 
square foot facility with approximately 17 people of whom 12 were 
professionals. (Palevsky, Tr. 3196, 3198; DX 45, p. 4.) Its first 
product was the SDS 910 computer system, delivered in mid-1962, less 
.- jj than a year after its organization. (PX 5774, p. 13.) That first 
j2 jj product was designed to take advantage of an opportunity perceived by 
, . ;j SDS for high performance hardware offered with little support to 

i g !i sophisticated customers for use in real time applications . The 

.i 

.,. \ market opportunity in fact existed and the 910 and subsequent 
lo 

i! products were very successful. SDS built on the success of its 
; ; initial specialization. Throughout the 1960s, it successfully 
;i expanded its product line both by offering its computers to a wider 
: set of customers for a wider range of applications and by producing 
: more and more of its own peripherals and software, which it had 
;; previously acquired from other vendors. 

22 " 



: * Palevsky defined a general purpose digital computer as "an 
*** ; electronic device with a stored program, which may be changed, and 

; depending upon the program, can operate a large variety of tasks". 
d5 , (Tr. 3132-33.) 

: ' -692- 



L 
2 
3 

4- 
f 

e 

7 
3 

10 

IX 
12 
12 
14* 
15 

10 

17 

IS 

19 



SDS grew at an extraordinary rate while also achieving 
substantial profitability — in fact, SDS "produced continually increas- 
ing profits virtually from inception". (PX 5774, p.. 6.) Its average 
annual compound growth rate from 1962 to 1968 was 115%. Even after 
the first two years, it continued to grow at a rate of approximately 
50% per year. (DX 46, summarizing data contained in DX 44 and DX 
45.) Its revenues, which by 1964 had reached $20.5 million, rose to 
$100.7 million by 1968, the last full year before it merged with 
Xerox. SDS was merged with Xerox in 1969, in exchange for Xerox 
stock valued at approximately $980 million. Of this amount, Palevsky 
received approximately $100 million worth of stock. He had also 
received several million dollars from previous sales of SDS stock. 
(Palevsky, Tr. 3195-96.) 

a. The SDS Entry Strategy. SDS implemented a consciously 
determined strategy to capitalize on what it saw as a market oppor- 
tunity. Palevsky testified that at its formation in 1961, SDS had 
"two markets" in mind for its products, "one market being what I 
would characterize as the real time computer market, and the other, 
the small to medium scientific computer market". (Palevsky, Tr. 



^ J 3133.)* SDS began to market computers of high performance hardware 
offered for real time applications to customers that did not need a 



21 
22 



* Palevsky described "scientific data processing" as processing 
where "a relatively small amount of data is entered into a computer, 
a large number of arithmetic operations are performed on data and a 
relatively small amount of data is produced in some printed form". 
"Business data processing", he said, "has the opposite meaning". 
(Palevsky, Tr. 3136.) 

-693- 



21' 

3 

A 

5 

>5 

.7 



lot of software and support services from the manufacturer.* 

(Palevsky, Tr. 3137; PX 2103 (Tr. 23290).) 

As; Palevsky testified: 

"It was; p»rt of, the- market, that essentially no one 
had attended to . At thdt time the other companies were 
really concentrating primarily on computers as devices 
into which one fed documents that contained data, cards, 
tapes, etc., and out of which one got printed answers. 

"Our computers were, intended for a market which fed 
real time data, that is, data that came from centers in a 
steam generating plant or a missile launching site or some 
astronomical instrument and produced signals that, say, 
worked the valves on a steam generating plant or indicated 
to other pieces of equipment within the launch site the 
status of various functions within a space vehicle so that 
it didn.' t work as a computer works in an air-conditioned 
computing center, but rather as part of the whole complex 
of operational equipment". (Tr. 3135.)** 

This strategy was highly successful. Palevsky testified 

that SDS , "at the beginning", was able to sell its products with "a 

very large gross profit" : 

"We were able to do that at the beginning because we provided 
hardware, that range of hardware and other services that was 



SDS did not initially attempt to market its product to "business 



.8 ■] data processing customers" because SDS "didn't have the kind of 
^| people who understood the business market and the need of the 

L9 i business market and we had not developed the software, the appli- 
1 cations engineering, the general support that the customer needed". 

iO !j (Palevsky, Tr. 3137.) 

i 

21 :| . ** In its 1966 Annual Report, SDS described its formation as follows 

;i 

22:1 "In 1961, when SDS was founded, highly experienced technical 

j personnel skilled in the design, production and marketing of 

23 [ small scientific and systems computers were uniquely available 

in southern California. During that same period, the scientific 
and engineering segment of the computer market required small, 

I real-time computers which could monitor and control experiments 

25 | or testing programs and rapidly process the results. Recog- 

1 nizing the requirements of this market, the initial objectives 

•j of SDS management were to attract competent technical personnel 

\ and effectively apply their experience to meet this demand." 

; ; (DX 982, p. 4.) 

I -694- 






L 
Z 

3 

4. 

ff 

7 
3 
? 

IQ 

LI 
12 
13 
1* 

Iff 



relatively unique and consequently the customer was willing to 
pay a relatively large sum for it." (Palevsky, Tr. 3155.) 

DEC was the only other firm Palevsky remembered producing products 

similar to those of SDS in the early 1960s. (Palevsky, Tr.. 3136.) 

The recognition by users of the potential benefits of the 

early SDS products — obvious from SDS ' s growing revenues — was expressed 

by NASA's Ames Research Center, which described its procurement of an 

SDS 920 computer system in 1963 as follows: 

"The integration of digital computers into physical systems 
dedicated to specific areas of research has only recently become 
economical and feasible through the reduction of equipment costs 
and component size. One of the first such systems in use at 
Ames was installed in 1963. This computer (SDS-920) was pur- 
chased by the Guidance and Control Systems Branch and applied to 
research on on-board computer and display requirements for 
spacecraft and aircraft*" (DX 5316, p. 9.) 

Subsequently, and rapidly, SDS expanded its product line and 

its marketing approach. SDS attributed its "early and sustained 

profitability" to its ability to meet the needs of its users: 



t ^ ., "Because of the rapid growth of the computer industry, the 

ic |j age of a company has not been a principal factor in its profes- 

:| sional or financial maturity. Far more critical in a company's 

-' »( potential is its ability to understand and act upon the changing 

' requirements of the marketplace. It is to this posture that SDS 

- s ;| conforms." (DX 44, p. 5.) 



i3 ;j b. The SDS 910. SDS ' s first computer was the SDS 910, 

2C ,} which Palevsky described at trial as a "special purpose general 

.1 
21 :l purpose computer" — by which he meant "a computer that had all the 

i 

22. :! characteristics of what was generally known as a general purpose 

23 ;i computer, with the added capability of operating . . . within a 

24. S systems environment, that is, it was a computer that was easy to 
i 

25 i integrate with diverse types of special purpose equipment". (Palevsky, 

i 

1 -695- 



Tr. 3132, 3134.) The "main frame" sold for $80,000 to $90,000? 
"[t]hen, depending on the peripherals, it got more expensive"-. 
Palevsky described the peripherals as being, "at the beginning, 
rather primitive equipment": paper tape punches, paper tape readers 
and card equipment. Also, at the beginning, the SDS 910 was marketed 
with "very primitive software, really just an operating system". 
(Palevsky, Tr. 3134,) 

c. The Expansion of the SDS 900 Series. In 1963, SDS had 
expanded its line by announcing the SDS 920, 930 and 9300. (DX 44, 
p. 7.) Those systems were compatible with the 910 and were designed 
with a "building block" design philosophy. ( Id. ) The 92 and 925 were 
introduced in 1964.* ( Id. , p. 3.) In its 1964 Annual Report, SDS 
told its stockholders that with the introduction of the "small, high- 
speed SDS 92 and the medium scale SDS 925, the company now offers a 
family of six compatible, general purpose computers — the SDS 92, 910, 
920, 925, 930 and 9300 — providing the flexibility required for both 
i| industrial and scientific systems". ( Id. , p. 7.) 

* SDS did not actually "manufacture" its 900 series com- 

ii 

li puters; rather it assembled them. That is, SDS purchased the various 



25 



* "The 925 was a modification of the 930 to . . . provide a faster 
lower-priced machine. It was software compatible, and was really 
just a modification of another product." (Palevsky, Tr. 3214.) 

-696- 



L I — Certain basic components for its central processing 

i 

Z '}■ units and memory (i.e., transistors, resistors and capacitors) 

3 I from the "fs]tandard avenues of supply — [f]rom the manufacturers 

4*1* of those components". (Palevsky, Tr. 3198-200.) 
3 \\ — Gore memories "at the beginning from Fabri-Tek" , which 

5 It was "one of a number of companies that supplied core memories"; 

7 S{ SDS acquired the memories in the form of core stacks, then 

1 

3 !| assembled them in boxes. (Palevsky r Tr. 3199.) SDS subsequently 
acquired core memories from Ampex, Magnetic Memories "and 
probably one or two others"; when "we got to a certain size we 
generally had three sources of supply so that we were always 
assured that one of them would be there". (Palevsky, Tr. 3200- 
01.) 

— Tape drives and tape control units from Ampex, Computer 
Froducts, and Potter Instruments; eventually, SDS made its own 
tape drives and controllers. (Palevsky, Tr. 3201-02.) 

— Printers and a few printer control units from NCR and 
Data Products and, in the case of "some specialized ones", from 
"small companies"; at the time of its acquisition by Xerox in 
1969, SDS was buying printer mechanisms from NCR. (Palevsky, 

I Tr. 3202; see Plaintiff's Admissions, Set I, 9 KH 191. 0-. 2.) 

i 
1 • 

J — Disk drives and disk drive controllers from Control Data 

(and "perhaps some of them from California Computer Products") . 

; (Palevsky, Tr. 320 3.) 
2± 1 

i 

"* i 

i -697- 



9 

ia 

LI 
12 

12 
14 ; 
15 
16 t 

is! 

» 

21 ;t 

22 



z 
1 

,4 
5 

J J 

L9 



— Card punches from Univac; card readers were initially 
acquired from a third party, but SDS subsequently built them 
itself. (Palevsky, Tr. 3203.) 

— k, few cathode ray tube terminals from Control Data; 
however , SDS built most of these itself. (Palevsky, Tr. 3203.) 

— " [Sltan&ard [Teletype] keyboard devices from Western 
Electric " . ( Palevsky , Tr * 3204.) 

SDS both wrote software for its computer systems and used 
outside software services. Software services were provided by 
Programmatic s and another firm, and "a number of smaller firms for 
very specialized things". SDS also had "a number of users groups and 
a number of our users ' programs became standard programs that were 
then widely distributed" by SDS. Additional software was obtained 
from a European company (a predecessor of CII) , which was licensed by 
SDS "to build our computers in France". (Palevsky, Tr. 3205-06.) 
SDS itself furnished maintenance service. (Palevsky, Tr. 3134-35.) 

The SDS 920 had certain instructions that were not included 
in the earlier 910 and "a slightly more sophisticated input-output 
system". It was marketed to essentially the same customers as the 
910. (Palevsky, Tr. 3162.) 



;| The SDS 930 was "larger and faster and, again, somewhat 

11 \ 

\ more complex structurally". It was partially marketed to the same 

21 ■:; 

j group of customers as the 910 and the 920, but was also marketed "to 

23 I 

S a greater extent to the general scientific community". (Palevsky, 

U I 

1 Tr. 3162.) For example, an SDS 930 was used for data reduction and 

mm 

i 

} -698- 



L 
2 

3 

4- 

3 

7 

a 

71 

12 
13 

14 
15 ; 
IS 



analysis at the Mississippi Test operations center associated with 
the NASA Slidell Computer Complex (DX 5836, Reeves, pp. 55-56) and, 
at the Kennedy Space Center, an SDS 930 performed off-line simulation 
of launch vehicle events for training, supplied input data to Mission 
Control in Houston and handled a "fuel loading" system. (DX 5652, 
pp. 116, 123, 164.) Palevsky testified that when it was first 
introduced, the SDS 930 competed with the IBM 1620 "[a]nd then when 
the 360 was introduced, the 360/30s, 40s and 44s". (Palevsky, Tr. 
3185.) Other competitors with the 930 were Computer Control Company 
(later bought by Honeywell) and DEC. (Palevsky, Tr. 3165.) 

Palevsky described the SDS 9300 as "conceived much more as 
a data processing system, as a computer that would sit in a central 
computing facility and essentially provide printed answers, as 
opposed to being interconnected on a real time basis with other 
sources of data". It was marketed "to the scientific community", but 
performed a still broader mix of applications. One customer was 



,_ ; DuPont, which had' previously integrated SDS computers "into systems 
:■[ for controlling chemical processes". It acquired this new computer 
I not only for a "specific process they wanted to control but rather 

19 i 

! for a general computing purpose, so that the customer may have been 
20 i 

J the same, but the part of the company would be different". (Palevsky, 



f i 

i 



! Tr. 3163.) Similarly, Digicon, Inc., used a 9300 to process seismic 
22. '*> 

t data collected from oil fields and also to process its accounting 

I records. (DX 4085, Poe, pp. 18-19, 21.) 
2± : ) 

! Palevsky testified that with the 9300, SDS "had now entered 

25 S 



. -699- 

i 



r 

3 ! 

.if. 

7 
3 
9 

a 

3 

L5 
La 
17 ! 



the more traditional and more highly developed market for computers 
and no longer had the edge of the innovations" it had made in "real 
time computers" on which to rely.* This engendered "[a] 11 the 
difficulties that go with, a highly competitive sales situation — 
marketing situation" \ (Palevsky, Tr.. 3164.) Palevsky testified that 
SDS's "main competitors' 1 * in marketing the 9300 were IBM and Control 
Data (and later, Digital Equipment). (Palevsky, Tr. 3165.) 

SDS had supported a "growing program of research and 
development" and committed "substantial capital to advanced product 
planning". One of the results was its announcement in 1964 of what 
it claimed was "the first computer to use monolithic integrated 
circuits, the SDS 92". As a result of the use of integrated circuits, 
SDS's manufacturing costs were "decreased while the reliability of 
SDS computers is improved at least three times over present models". 
(DX 44, p. 5.) Withington, writing in 1964, concluded that: 



"The most significant development in components has been the 
approximately 50% reduction in the manufacturing cost of high- 
speed circuits over the past three years . This quite rapid 
development has enabled new small companies (e.g., Scientific 
Data Systems, Digital Equipment Corporation) to enter the com- 

-3 ; { puter market with low-priced computers of high performance. . . . 
|j This reduction in manufacturing cost has been at least partly 

IS \ responsible for the recent price reductions on older computers 

! and the lower prices of new ones. The user has benefited, and 

2G j the market has been enhanced." (PX 4829, p. 31.) 

'I 

21 j By the end of 1964, SDS told its stockholders that its com- 

i 

22 ; puter systems were "presently being used by industrial, scientific and 

£-3 ' 



24. I * "As the technology in the computer industry evolved, there were 

i no longer those pockets , there were no longer those market areas that 
25 j had relatively little competition." (Palevsky, Tr. 3155-56.) 

j 

"] -700- 



L 
2 



government organizations in many diverse applications ranging from 
space exploration to construction, medical research to food process- 
ing". (DX 44, p. 3.) The number of SDS employees had increased from 
438 in 1963 to 1,357 at the end of 1964. ( Id. ) The year 1964 was 
also a year of expansion abroad. According to SDS's Annual Report for 
that year, " [f]rom nuclear experimentation in Geneva to automotive 
manufacturing in Tokyo, SDS computer systems are finding an accelerat- 
ing and receptive market throughout the world. . . . In the first 
significant year of SDS activity abroad, computers were ordered or 
installed in more than 15 countries." ( Id. , p. 17.) 

Expansion, plus SDS's program of research and development, 
required capital. Requirements for capital also increased because, 
according to Palevsky, SDS was leasing more of the 9300s than it had 
prior computers. (Palevsky, Tr. 3164.) In 1964, "due to [its] rapid 
growth", SDS made its first public offering of common stock, offering 
Tg \ 382,375 shares and raising almost $5,000,000. (DX 44, pp. 3, 21.) 

SDS engaged in a continued pursuit of growth and expansion 
through continued product improvements. In SDS's 1965 Annual Report, 

1Q :! following the announcement of IBM's System/360, SDS stated: 

•t 

;| "The character of the computer market changed substantially 

^ j last year as the result of advances in both the understanding of 

the technology and in the manner in which computers should be 

employed. . . . [T]hese changes point to the increasing use of 



4 
4-1 

5 

7 

3 

9 

10 

n 

12 

i 

14- 1 
15 I 



17 



■\ 



18 



7f 



22 * 

I 

23 1 



| total management information systems for business, scientific, 



aerospace and industrial control applications. 



"As is always the case in the computer field, the new 
market demands increased performance economically, in terms 
2- ; of more computations per dollar, and an expanded array of 

supporting services such as programming, field services and 



/*s ! 



^ \ training." (DX 981, p. 4.) 



-701- 



L 
2 
3 

4- 

3 



In 1965, SDS brought out the 940, designed for simultaneous 
access by multiple users at remote locations. SDS took the 930, 
increased its memory capacity, and integrated rapid access data 
storage units and communications equipment. A 9 30 system costing 
$250,000 was thereby transformed into a 940 system costing $1 million. 
(DX 981, p. 3; DX 982, p. 12.) SDS called the 940 a "timesharing 



7 |l computer". It was used, among others, by several commercial time- 



sharing service bureaus (DX 45, p. 7) as well as by a data center 

established by SDS itself to sell time to remote users in the Los 

Angeles area. (DX 983, p. 3.) Similarly, White, Weld & Co. used its 

940 to implement a financial information system that permitted its 

individual subscribers to request portfolio information on a variety 

of companies. (DX 982, p. 12.) 

By the time of the introduction of the 940, SDS had announced 

the development of its own magnetic tape units and rapid access disk 

.£ ; i files as well as a line of digital logic modules. (DX 981, p. 3.) 

,„ '} Also, in mid-1965, SDS "announced a new business programming* 

it package for all its computers to supplement the extensive library of 
- s ;l 
.- !{ programs presently available to scientific users". The package, known 

i 

;t as MANAGE, was "expressly designed to facilitate corporate decision 

20 j 

l making bv management personnel outside of the data processing depart - 

21 ■! • 

|i ment" . (DX 981, p. 4.) Similarly, SDS adopted some of the marketing 
22. 'I 

\ practices of others in the industry. In 1965-66, it offered the 
23 i 

1 Federal government a 14% discount towards equipment purchase " [f]or 

2* 1 

! qualified Government schools and training institutions when primary 

25 :j 

i 

-702- 



3 

10 

II 
12. 

12 
14- 
15 



I 

2 

3 

4- 

5 
a 
7 ; 

a 

9 ; 

ia 

II 



application of the data processing system is for educational and 
training purposes". (DX 47, p. 16.) It also offered the government 
the provision of "programming aids , including programs [ , 3 routines , 
sub-routines, translation compilers and related items without extra 
charge". ( Id. / p. 3.) By 1966 SDS was marketing its computers on a 
variety of lease terms as well as selling them. (Palevsky, Tr. 3207.) 
SDS continued to grow dramatically during 1965. It doubled 
its number of installations in one year. New business received during 
the fourth quarter of 1965 was greater than any previous quarter in 
the company's history. To finance this continued expansion, SDS sold 
47,500 shares of $100 par cumulative preferred stock to a group of 
y* j insurance companies, raising $4,750,000 in the second quarter of 1965. 
22 | And in February 1966, it sold $10 million in convertible subordinated 
• . j debentures to a group of institutional investors. (DX 981, p. 3.) 

d. The Sigma Series. By 1965 SDS had begun the develop- 
ment of its Sigma series of computers — its third-generation line. In 

jjfact, in that year (following IBM's announcement of System/360), the 

17 i t 

;[ Sigma family "occupied the attention of virtually every department in 

:i 

the company". (DX 981, p. 9.) It was announced in 1966, with the 



15 



13 ; 

20 ' 



first of the series, the Sigma 7, announced in March of that year and 



lithe second, the Sigma 2, in August. The remainder of the family was 

21 -j . 

j announced starting in 1967. (Palevsky, Tr. 3226.) By 1971, the Sigma 
22 



23 
25 



family included the 2, 3, 5, 6, 7, 8 and 9. (PX 5774, p. 13; DX 
13400, p. 22; DX 13401, p. 19.) According to SDS, the Sigma family 

delivered "more computations per dollar than any other commercially 

i 

1 
I 

ij -703- 

•i 

i 

■j 



1 ; 

i 

3 i 

+i 

5 ! 

i 

5 : 

i 

7 I 

t 
1 

3 J 
9 ! 

10! 
12 i 

13 i 



T4>> j' separate boxes at prices below IBM's, but a comparison could not be 



15 



17 ! 



available machine". SDS touted its versatility: "The impact of 
Sigma, however, lies in its broad application for business, industry 
and science as well as its ability to perform an almost unlimited 
number of different applications at virtually the same time." (DX 981, 
p. 6.)* 

The Sigma Series was a response to the IBM 360 line. A 
Sigma 7 press release, dated March 15, 1966, stated that Sigma "repre- 
sents the first family of computers with an entirely new design since 
the IBM 360 announcement" and that "Sigma 7 features a total capability 
for both business and scientific data processing". (DX 52, p. 1.) 
SDS attempted to set its prices 10 to 15 percent lower than IBM's 
prices on the products that IBM had announced two years earlier. 
(Palevsky, Tr. 3150.)** An effort was made to price each of the 



made on the basis of the performance of the systems as a whole because 
1C ! "then you get into the problem of what is a typical set of operations 
jl and it becomes very complex to do". (Palevsky, Tr. 3269-71.) Appar- 
it ently, SDS felt that a dollar price advantage was necessary to over- 
'i come the obvious customer acceptance of System/360. (See Palevsky, 

19 ;j 

_ j Tr. 3149-50, 3176, 3270-72.) 
20 ;! 



21 || = ■ 

|j * For example, SDS contended that a Sigma "can simultaneously run 
^2 ij an inventory control program together with a real-time process control 

i| application. At the same time, 200 users at remote consoles through- 
^ lout the country could be time sharing the central processor". (DX 981, 

24 l 

^ .« ** Currie testified that "generally" although "not always", SDS j 

^ ;| tried to have somewhat lower prices than IBM for equivalent perform- 

• ance on the order of 10 to 15%, and tried to have "an advantage over a \ 

i company like Univac" . (Tr. 15175-76.) ! 

;| I 

'i i 

I ! 

;j -704- 



L I Palevsky testified that the Sigma computers compared to the . 

Z <j 9300 "were more complex structurally. They were much faster, and some 
3 ; of the computers in that line were compatible so that we had on a very 
^\\ small scale something like the 360, that is , we had a number of 

computers of various sizes that were program compatible," (Palevsky, 
5 ; l Tr. 3165-66.) As with the 360, SDS "designed standard interface 
7 j| units". It also "developed special programs which simplify the 

design, engineering, and final assembly of various building blocks 
or components into total systems." (DX 982, p. 12.) 

For the Sigma series, SDS acquired Potter tapes, Control 
Data disk drives, NCR printers, Teletype console typewriters and 
Uptime card equipment* (Currie, Tr. 15507.) However, because periph- 
eral equipment was viewed as a "critical element" in third-generation 
computer systems, SDS had sharply increased its planning for internal 
development and production of peripherals. In 1966, SDS began deliver- 
ing its own magnetic tape units and had completed development of Rapid 

;; Access Data files, which it called "two important peripheral products 

17 i 

for data storage that were completely designed and produced by the 

i 

j company". These products were expected "to enhance the capabilities 

I of SDS computer systems". (DX 98 2, p. 10.) SDS stated its reasons 

:i for undertaking peripheral equipment development programs internally 

21 \ 

\ rather than acquiring independent manufacturers or continuing to 

22 ''» 
j purchase from suppliers as follows: 

22 i 

[ "First, SDS can realize a significant improvement in profit 

24. 1 margins on equipment which the company produces internally. 
! 

25 } 

j 

:! -705- 



3 
9 

ia 
li 

12 

13 

14- 



13 
IS 



"Second, peripherals designed to complement the capabilities 
of Sigma computers provide an additional competitive advantage 
for SDS systems. 

"Third, high reliability must be designed into the equipment 
and quality control assured during production, thus minimizing 
the cost for servicing faulty peripherals. 

"Fourth, and most important, the technology is advancing too 
rapidly to permit SDS to rely primarily on suppliers. The new 
series of Rapid Access Data files is an example. Developed with 
a considerable investment., SDS RADs are among the most advanced 
secondary storage devices in the industry. The availability of 
the various RAD models provides SDS with a significant advantage 
in marketing Sigma computer systems." (DX 982, p. 10; see also 
Palevsky, Tr. 3277-78.) 

SDS provided "advanced software, including operating systems 

for real time , batch , and time-sharing operations , FORTRAN IV and 

COBOL compilers, assemblers" and various applications software, 

including a library with "[m]ore than 1000 utility and mathematical 

programs" for the Sigma family. (DX 49, pp. 2, 8.) SDS obtained 

between 20% and 50% of this software — specifically assembly languages, 

compilers , a Data Management System package , a linear programming 

5 |j package and a communications package — from software houses such as 

7 ij Digitek, Programmatics , Bonner & Moore, Informatics, Computer Usage 
:l 

3 .j Corporation, Computer Sciences Corporation, Dataware, CII , and 
9 j Scientific Resources. (Currie, Tr. 15388-89.) 

6 | At first, according to Palevsky, the Sigma series was 

'1 | offered "to essentially the same market as before". Gradually SDS 

4 

2. '.I "started to market it for applications that were mixed scientific and 

2 ; general data processing" (Palevsky, Tr. 3166) , and began to expand 
! 

It "\ into applications for general business and industry, including market- 
.1 

\z j ^ n< ? to business data processing customers. (Cohen, Tr. 14684-86.) By 

i 

i 
i 

I -706- 



L 
2 

3 
4. 



this time, "changes in the technology" had begun to "blur" the dis- 
tinctions between business and scientific data processing. (Palevsky, 
Tr. 3137.) More and more customers started using a single computer 
■ for both types of computation*; if one were to ask "say in '65 how 
many installations used a single computer for both purposes and say by 
'68 how had that grown ... I would guess that that had grown ten- 
fold". (Palevsky, Tr. 3254-55.) Both because of that change and a 
perception that the area SDS was focusing on was becoming too confin- 
ing, SDS was "forced" to enter the "market" for business data process- 
ing customers. (Palevsky, Tr. 3137.) 

At the time of the Sigma 7 announcement, SDS issued a press 
release stating: 



3 

7 i 

3 

9 

ia 

u i 

12 
12 

L4- 
15 

IS; 

17 j 

IS ( 

; | (DX 932, p. 3.)* 
IS 

\ SDS was advertising its Sigma 7 computer system as "unfair i 

20 I 

1 to IBM . . . Sigma 7 does everything a 360/50 does. At a fraction of j 
7f ! i 



"Until now, explained SDS president Max Palevsky, computers 
were generally built either for business data processing or for 
solving problems of a scientific nature or for real time control 
systems. 'But because of its advanced internal architecture,' 
Mr. Palevsky stated, 'Sigma 7 is the only medium priced computer 
that can deliver outstanding performance in any of these applica- 
tions. ' " (DX 53, p. 1.) 

Similarly, the Sigma 5 was "designed for operating real-time programs 

■ simultaneously with general purpose scientific and business problems". 



22 ■ 

| * In its 1966 and 1967 Annual Reports, SDS stated that although its 
2- i Sigma series would perform business data processing applications, that 

J did not mean that it was abandoning the customers on whom it had 
25 heretofore built its business. (DX 982, p. 4; DX 983, p. 4.) 

-707- 



4 the cost. Sigma 7 is a little cheaper than the 360/50 and a good deal j 

22. '"'» • ! 

i i 



u 
r ! 



faster. The combination gives Sigma a 25 to 65 percent edge in 
cost/performance." (DX 54.) Cohen testified that the Sigma 7 "did 
indeed" perform business data processing applications. (Cohen, Tr.. 
14631.) For example, the Harrison Radiator Division of General Motors 
used Sigma equipment for its data processing requirements, which were 
"principally" inventory control, which, according to Cohen, would be a 
T jl business data processing application. (Tr. 14610-11.) Cohen also 
I : \ testified that, looking only at the CPU, input/output processors and 
5. ; main memory, the Sigma 7 could in fact do everything that the IBM 
3 I 360/50 could do. (Cohen, Tr. 14622-24.)* 

1 | According to an XDS sales guide (DX 50, p. 102/001-29), a 

? 1 report based upon what was happening in the field, the 360/50, 360/40, 
44, 65 and the 1800 were "competitors" of the Sigma 5 as late as 1972. 
(Palevsky, Tr. 3232-33; see also Tr. 3185, 3228-29.) According to 

Cohen, IBM's 360/44, 50, 65, 67 and 75, as well as the 1130 and 1800, 

t| 

^ !i were the IBM systems SDS "most commonly competed with". (Cohen, Tr. 

■ 
7[ 



3 



14555-56; PX 433.")** 



ii 



!l * However, when the array of peripherals and software available on 

- I the IBM 360/50 was taken into account, Cohen was of the opinion that 
.1 the Sigma 7 could not do all of what the 360/50 could. (Cohen, Tr. 

•^ •] 14624.) Palevsky testified that the Sigma 7 was comparable to the 
I 360/50 in terms of the hardware capability, but not in terms of the 

- ! total system. (Tr. 3243-45.) 
■j 

!2 ii ** j n the early 1970s, the "prime competition" for the Sigma 5 was 
•] the IBM 370 Models 135, 145 and 155; the DECsystem 10; the SEL (Systems 

2 '■ Engineering Laboratories) 86/3 8 and the Univac 418-III. (Palevsky, 
! Tr. 3231-32; DX 50, p. 102/001-29.) For the Sigma 6 and 9, the "prime 

•- ] competitors" were the I3M 370 Models 135, 145, 155; the DECsystem 10 
and the PDP 10, Models 1040, 1050, 1055 (dual processor); and the 



?c 



j Univac 1106. (Palevsky, Tr. 3247; DX 51, p. 103/001-13.) For the 



-708- 



z 

4* 



a 
7 
S 
9 

iai 

Ll 
12 

13 
14 
15 

15 i; 
17 I 

is ; 

19 
ZQ 



Competition for the Sigma series included products from many 
suppliers in addition to IBM. Cohen listed 21 companies (plus leasing 
companies) as competitors to SDS/XDS "in the computer systems market" 
during the period 1966 to 1972: IBM, Honeywell/ Univac, GE r CDC, 
Burroughs, DEC, SEL, Modular Computer Corporation, Fischer & Porter, 
Varian, Hewlett-Packard, Data General, Radiation, Inc., Harris, 
Collins, Comten, Interdata, Electronic Associates ("on occasion") , 
EMR, RCA ("occasionally") and said there were "very likely" others. 
(Cohen, Tr. 14600-09.) 

Palevsky mentioned as competitors to SDS : IBM, CDC, DEC, 
CCC (Computer Control, later acquired by Honeywell) , Univac ("occa- 
sionally") , Burroughs ("rarely") and Honeywell (rarely, until the 
latter' s acquisition of CCC) , and General Electric ("in certain 
applications"). (Palevsky, Tr. 3166.) Engineering Associates, CCI, 
NCR, ICL, EMR and COM were also competitors. (Palevsky, Tr. 3233; DX 
50, p. 102/001-30.) 

Cohen listed the "effective competitors" (he defined these 
as companies which won 20% to 25% of the competitions in which they 
;jwere engaged) (Cohen, Tr. 14723-24) of SDS in specific application 

•.i 

\ areas • 



i In time sharing, he listed IBM, Honeywell (after the acqui- 

j sition of General Electric 's computer business) , Univac, GE and 



iSigma 8, the "prime competition" was the IBM 370 Models 145 and 155, 

24 Ithe SEL 86/88, DECsystem 10, the Univac 1106/110 8, and the CDC 6200/ 
|6400. (Palevsky, Tr. 3238; DX 50, p. 102/001-30.) 

25 j 

-709- 



) 

! 

l \ 



DEC. 

In real time, he listed IBM, Honeywell, Univac, CDC, 
DEC, SEL; 

In seismic, IBM and> Univac? 

In scientific batch, IBM, Univac, GE and later Honeywell 
(after its acquisition of GE's computer business), CDC and DEC; 
In communications, GE, Univac, IBM and Comten; and 
In multi-use/multimode, IBM, Honeywell (after the GE acqui- 
sition) , Univac, CDC and DEC. (Cohen, Tr. 14729-30.) 

IBM competed with SDS as an "effective competitor" in every 
application area, and the systems which competed with those of SDS 
spanned the IBM product line: the 1130, 1620, 1800, 360/44, 360/50, 



.•[360/65, 360/67, 370/145, 370/155 and 370/158. (Cohen, Tr. 14555-60; 



.*! 



7 
3 



PX 433.) 

Indeed, Wright — IBM Director of Time-Sharing Marketing in 
the period from 1964 to 1965 and Director of Marketing for Government, 
|| Education and Medical Region from 1965 to 1969 — testified that SDS 
; was among IBM's "principal competitors". (Tr. 12993.) IBM was well 



; | aware of SDS during the mid- and late-1960s. On December 22, 1964, 
"I Lear son reoorted to T. J. Watson, Jr., about the serious competition 
i SDS posed to IBM's 360/40 and 360/50. (PX 1288, p. 2.) That concern 
; intensified over the next several years. (See the discussion of the 



?? 



£d 



Model 44, pp. 412-13.) SDS also appeared as one of the nine "major 
computer manufacturers" reported on in internal IBM reports on the 



;-j financial results of certain of its competitors. (See, e.g., PX 

IS j 

~ 13451.) 

;t 

;| -710- 

! 
■i 



L 
Z 
3. 
4* 

6 

7 



a 

10 

II ! 



e. The Merger. The Sigma family brought still more growth 
for SDS. During 1966, which was described by SDS as "a crucial 
transitional year", SDS offered two issues of convertible subordinated 
debentures totaling $22,500,000 and retired short-term bank loans. 
(DX 982, p. 3,) From, the fourth quarter of 1966 to the fourth quarter 
of 1967, the year which SDS saw as a "critical" period "during which 
the character of . . . future expansion" was "largely determined", SDS 
doubled its production of EDP equipment. It stated that in 1967, it 
"successfully completed its first product line transition, began a 
major facilities expansion,, and initiated new product development and 
cost control programs to sustain orderly growth". (DX 983, p. 3.) 



££ j By 1969 , SDS had achieved that orderly growth and had reached 



13 
14- 
15 
IB 



what would be, for SDS as an independent entity, the pinnacle of its 
success . It told its stockholders that its international sales in 
1968 had increased by more than threefold over 1967 and that "[t]he 
sale of SDS oroducts outside the United States is expected to continue 



,- !| to increase significantly through the company's increasing involvement 

\ 
1S J in the various international markets". (DX 45, p. 12.) SDS stated 

that it "ranks among the world's ten largest computer manufacturers" 



| with assets of $113 million and more than 4,000 employees ( id. , p. 4), 



19 

I and described itself as "one of the world's largest suppliers of 

ZL \ ■ 

•I commercial time-sharing systems". ( Id. , p. 6.) 
22 :» 

In 1969, SDS was acquired by the Xerox Corporation, a 
company that had achieved "a position of eminence as a worldwide 
;! enterprise" through the remarkable acceptance of the xerographic 

:! -711- 



L 
r 

r\ 

i 

I \ 

3-1 

3 i 

! 

3 ! 

.4. : 



3 '« 



7!! 
si 

•i 



2. 1 

•t 



copier. (PX 5774, p. 6.) Xerox's revenues for the year 1968 (exclud- 
ing Rank Xerox, Ltd., its British affiliate that marketed its products 
abroad) were $896 million and its net income was $116 million (includ- 
ing the income from Rank Xerox). (DX 13857, p. 3.) SDS was acquired 
for approximately $980 million worth of Xerox stock.* (Palevsky, 
Tr. 3195.) 



[c j * The Xerox acquisition of SDS and the subsecuent activities of 
) XDS are discussed below. (See pp. 1125-44.) 

i 

! -712- 



1 

2 1 



*■; 

a 

7| 
3 

id 

n 

12. 
13 

I* 

la ;i 
17 



48. Digital Equipment Corporation. Digital Equipment 
Corporation ("DEC") was founded in 1957 by Kenneth Olsen and 
Harlan Anderson (Hindle, Tr. 7318.),* both having previously been 
**;' associated with M.I-T.' s Lincoln Labs where they worked on 

Whirlwind and SAGE. (DX 13833, p. 5.) dec set up production 

on one floor of a converted woolen mill in Maynard, Massachusetts, 

with three employees. (DX 13858, p. 1.) Its initial capitalization 

was $70,000, all of which was invested by the American 

Research and Development Corporation — a Boston-based venture 

capital firm. (Hindle, Tr. 7476.) Its first products were 

laboratory logic modules — "printed circuit boards containing 

components which are used to do logical functions in an 

electronic sense: add, etc." — that were then used to test 

and build other manufacturers 1 computers. (Hindle, Tr. 7318-19; 

DX 13858, p. 1.) 

The story of DEC is one of extraordinary growth 
and enormous success in the computer business. From its 
.- j beginning and throughout the 1960s, DEC achieved extremely 

19 '.j j 

„_ i * Winston R. Hindle, Jr. was the only witness from DEC. j 

^° j Mr. Hindle joined DEC in 196 2. From 1967 through j 

„ t : | his testimony in 1975, Mr. Hindle was Vice President and j 

^* :{ Group Manager of the company with responsibility for numerous j 

■j products within the DEC product line. His responsibility j 

2— j encompassed the development, marketing, sales support, ; 

j planning and financial areas. Mr. Hindle had served on j 

— s j the executive committee of DEC, as well as the Finance and j 

| Administration and Marketing Committees. (Hindle, Tr. j 

24 } 7313-18, 7337.) As of 1979, Mr. Hindle was Vice President, | 

^ j Corporate Operations. (DX 12323, p. 47.) 

25 ■ • ! 

1 I 

1 -713- 

•;t ! 



t ! rapid growth by taking advantage of new technology and its 
* I own research and development to manufacture an ever-expanding 
3.! product line. DEC'S total assets grew from $5.7 million in 
*|.: mid-1964 to $114. 8 million' in. mid-1970 . * (DX 511, p. 14; 
5 DX 1384-S, p. 10.) DEC's profits after taxes went from $889,000 

in 1964 to $14.4 million in 1970 (DX 511, p. 1; DX 13845, p. 10.) 

Its worldwide EDP revenue grew from $4.3 million in 1961 to 

$12-6 million in 1964, to $142.6 million in 1970, rising at 

an annual compound growth rate of 44% per year. (DX 526.) 

By virtue of the $70,000 investment in 1957, American Research 

and Development had acquired 78% of DEC's common stock. 

(DX 13833, pp. 5-6, 21.) In 1968, it sold 215,000 shares of 

DEC stock for a gain of more than $26 million (DX 13834, p. 9)/ 

and in 1972, it distributed the remainder of its DEC stock, 

valued at $382 million, to its shareholders. (DX 514, p. S; 

DX 13835, p. 4.) 

DEC acquired more and more space in its woolen 
mill, and its original three employees were joined by many 
others. It began expanding overseas in 1964. It formed 
its first sales subsidiaries in the United Kingdom and 
Australia in that year. In 1965, offices in Canada and 
Germany were added. (DX 13845, p. 3; DX 13846, p. 3.) Sales 



* Financial information for DEC first became publicly 
available in 1964. 



-714- 



I 

2 

Z 



offices in France, Japan and Sweden followed in 1966. (DX 13847, 
p. 19.)* By 1966, DEC occupied 338,000 square feet in its original 
location in Maynard, Massachusetts and employed 1100 people (DX 13847, 



4» ! p. 3), had 24 sales offices in six countries, and had about 800 



3" 

e 

7 

a 

9- 
10 

II 
12 

13 
L4 
15 
IS ,. 

i 
is I 

19 I 

20 ! 



computer installations. (DX 517, p. 1.) By 1970 it had manufacturing 
plants in England, Puerto Rico and Canada, as well as several in 
Massachusetts, employed 5,800 people (DX 511, p. 3.), and had 
computer installations in eleven countries. (DX 517, p. 2.). 

The financing of this expansion required capital. In 
1963, $300,000 was borrowed from American Research and Development 
Corp. However, by the end of its 1964 fiscal year, DEC had accumu- 
lated over $3 million in retained. earnings. (DX 13845, pp. 10-11.) 
Retained earnings rose to $4.3 million in 1965, $15 million in 
1968, $24 million in 1969 and $38.8 million in 1970. (DX 511, p. 15; 
DX 13846, p. 12; DX 13979, p. 7.) DEC made its first public stock 
offering in August 1966, raising $4,800,000. (PX 5026, p. 15;** 
DX 13847, p. 3.) From 1968 to 1970 it had three additional public 
offerings, raising a total of $63.5 million. (PX 4562, p. 17; DX 
511, p. 17; DX 512, p. 11; DX 13979, p. 8.) A review of DEC'S Annual 



£t j * In 1961, all revenue was domestic. In 1964, 91% of revenues 

|j were generated domestically. By 1970, the domestic percentage had 
22? dropped to 72%. (DX 526.) 



23 I 
24 

25 



** A more legible copy of PX 5026 has been marked as DX 13848. 



-715- 



I 
z 
a 



Reports reveals that this outstanding record enabled DEC to meet 
its financing needs with no reliance on long-term bank loans. As 
Hindle testified, "Digital's expansion has not ever been limited by 
the ability to raise cap ital . " (Tr. 7476.) 

It is interesting to note in this connection that DEC 
generally chose not to tie up its capital in financing leases for 



7 | r its customers, feeling that it had "better ways to invest" its 

SI money. When its customers wanted to lease, DEC would put the 

9- j customer in touch with leasing agents who would "use their capital 

10' I. and not Digital's capital". These "leasing agents" were organizations 

11 .; "willing to finance a customer's computer over a period of time". 

12.! When DEC did offer leases directly to its customers, it generally 

i 
13 sold them to financial institutions immediately thereafter. 

■ 

IA, (Hindle, Tr. 7369-70.) 

15 i DEC'S success was due in part to its commitment to product 

I 

15 |f development. Hindle testified that "Digital has through the years 

i 
i 

U | spent between 8 and 11 percent of revenues annually on research and 

IS I product development. ... We have felt that product development 

i 
ig • was a vital part of our success. The rapid advances of technology 

20 | in the computer field have meant that we must keep abreast of these 

2i | advances and incorporate them and understand them in our product 

22 H lines as rapidly as we could." (Hindle, Tr. 7383-84.) That 

23 |{ commitment paid off and, as Withington testified, DEC "always 
24. ■ maintained a position of technological leadership or at least 
25 ; currency with any significant competitor and always provided an 

-716- 



I 

2 

3 

4* 

5 

&i 

7 
3 

9* 
10 

II ! 

12! 

1A ! 
15 

16 J 

17 ! 
IS j 
19 



adequate breadth of product line, maintenance and support". 
(Withington, Tr. 56016.) 

DEC's computer systems began with small, high performance 
hardware offered with relatively little support to sophisticated 
customers capable of providing themselves the software and services 
for the application of those products to their needs. Palevsky of 
SDS described the early DEC computers as being similar to the small 
and medium-scale 900 series computers marketed by SDS principally 
for real-time and scientific applications, also in the early 1960s. 
(Palevsky, Tr. 3133-36.) Over time, DEC's products grew in capacity 
and capability and DEC expanded its customer support, its marketing 
and its software and service offerings . That process has continued 
and DEC now offers one of the broadest product lines in the industry 
and markets it to the whole spectrum of EDP customers. (See below, 
pp. 717-31, 989-92.) 

DEC's first computer, the PDP 1, was delivered in 1960, 
and it was followed by the PDP 4 and 5 in 1962 and 1963, respectively." 
From 1964 through 1970, DEC also introduced the PDP 6 (1964), PDP 7 
!l (1964), PDP 8 (1964), PDP 8S (1966), LINC 8 (1966), PDP 9 (1966), 



„„ ! PDP 10 (1967), PDP 81 (1968), PDP 8L (1968), PDP 12 (1969), PDP 14 
20 j, 

I (1969), PDP 15 (1969) and PDP 11 (1970). All of these DEC computers 

»• it ! 



i were classified by Hindle as "general purpose computers", except 



23 ! 

24 ! 
25 



the PDP 14.* (Hindle, Tr. 7321-24, 7327, 7388; PX 377-A. ) As we 



i 

* The PDP 14 was not a general purpose computer "because it has I 

a program which is preset prior to its delivery, and then it only ; 

operates on that same program when used by the customer". (Hindle, I 

Tr. 7327.) j 

i 
i 

-717- 



I 

2 

Z 

4- 

5 

6 

7 

3 

9- 
10 
11 
12 
13 

15 i 

16 

17 

IS 

19 



shall see, each of DEC's successive product announcements expanded, 
the breadth and capabilities of its product line, contributing to 
DEC's phenomenal success. 

a * PDF 1/ 4, 5 and 7. DEC's first computer was the PDP 
1, delivered in 1£6Q (and withdrawn in 1963 or 1964) . (Hindle, Tr. 
7318-19, 7321; DX 507, p. 10.) The PDP 1 was "an outgrowth of the 
technology that was incorporated into the line of logic modules , 
although there, was a completely separate design from the logic 
modules in our product line prior to that". (Hindle, Tr. 7319.) 

The original purchase price of a PDP 1 was in the neighbor- 
hood of $125,000 to $243,000. (DX 13858, p. 2; Plaintiff's Admissions, 
Set II, 1111 240.2, 371.3(d).) It was designated "PDP" (Programmed 
Data Processor) , a nomenclature that DEC used throughout the 1960s 
for its products, because "EDP people could not believe that in 
1960 computers that could do the job could be built for less than 
$1 million". (DX 13858, p. 2.) 

At Stanford University, Professor John McCarthy used a 
PDP 1 to conduct some of the earliest research on time-sharing in 
the early 1960s. (Feigenbaum, Tr. 29531-32, 29535-36; DX 13858, p. 



20 :' 3.)* A PDP 1 was also used, in conjunction with an IBM AN/FSQ-32 

2i |i 

22 i; * The PDP 1 "operated primarily independently", but was "linkfed] 
| through a disk" to an IBM 7090. " [O] ccasionally there would be a 

23 ;! lash-up which would exercise LISP on the 7090, which is a system 
developed ... [at Stanford] that would communicate through the 

^ disk to the PDP-1." (Feigenbaum, Tr. 29532-33. ^ 

-718- 



1 ; | computer, as "the major input/output vehicle for the various remote 

2 J devices" in a "general -purpose time-sharing system" at the System 
3-i Development Corporation (SDC) in June 1963.* (DX 7622, p. 3.) 

4 I That time-sharing system was produced "under the sponsorship of 

5 | ARPA[**] and. . - • utilized ideas developed at both Massachusetts 
$ !i Institute of Technology and Bolt, Beranek, and Newman, as well as 

7 i| some original techniques". The various remote devices used as part 

3 ;! of this system included "Teletypes . . . and other computers" that 

g I) could be run "from within SDC, and from the outside". PDP l's, as 

i 
1Q i well as the CDC 160A and the IBM 1410, were also expected to be used 

it 
H J at remote stations as part of this system. ( Id. ) In 1962, the 

■i 

12 |l Atomic Energy Commission's Lawrence Livermore Laboratory selected a 

ii 

13 || PDP 1 over a 1401 proposed by IBM to perform what it described as 

ii 

14 i "scientific and engineering" applications. (DX 2992, pp. 18, 

ii 

15 ;! 1113.) 

15 : | DEC introduced the PDP 4 computer system in 1962, and the 

17 l| PDP 5 in 1963. (Hindle, Tr. 7321.) The PDP 5 was offered with 

,a 1 keyboard-printer, paper tape reader and punch and a software package 

19 ! 

i 

2q 1 * "Time-sharing, in this case, means the simultaneous access to 
i a computer by a large number of independent (and/or related) users 
2i ! and programs." (DX 7622, p. 3, emphasis in original.) 

22 



** ARPA is the government's Advanced Research Projects Agency, 
which was established in 1958 and whose, "primary mission ... is 
2- ■; to support research and development of advanced projects which have 
potential value to the Department of Defense". (Plaintiff's 
Admissions, Set I, 1MI 1.0, 2.0.) 



±3 
24 
25 



-719- 



i 
f 

z 

T 
3 

q : 



including FORTRAN. (DX 13928 .) "Although the term •minicomputer' 
wasn't used during the time of the PDP- 5," DEC would have considered 
the PDP 5 a "minicomputer". (Hindle, Tr. 7325.)* According to the 
1964 DEC Annual Report, the- PDP 5 was used in "numerous applications 
in physics , biomedicine, industrial process control, and systems 
applications " . (DX 13 845 , p . 8 . ) 

In the early sixties, DEC'S research and development 
effort produced the "flip chip" modules (DX 13845, p. 3) which, 
like IBM's SLT, combined printed circuits with discrete components 
( id . , p. 2) and which made possible the introduction in 1964 of the 
smaller, faster and cheaper PDP 7 and 8 to replace the PDP 4 and 
5.** An IBM Win and Loss Report for August 1964 reported competition 
between the DEC PDP 4 and the IBM 1401 and between the PDP 7 and 
^ : . the 360/30. (PX 3630, p. 6.) The Atomic Energy Commission selected 

a PDP 7, installed in January 1967, over a 360/30, as well as systems 

i 

bid by CDC, SDS, Univac and Honeywell. (DX 2992, p. 49.) 

.a ■ 

j ;| 

[\ * Hindle testified that the term "minicomputer" (which "[pleople 

•S :j in the industry started to use ... in the middle 1960 's") "is not 

;j a precise term", but rather has "a broad range of definition [s] " as 
•S :} used in the industry. Hindle* s "own view" of a minicomputer is a 

j system "priced at less than $50,000". (Tr. 7325.) By this he 
£G j meant that "the smallest available configuration could be con- 

:j figured for less than $50,000". (Tr. 7453.) 
U ■! 

j ** Withington commented in 1964 that the rapid reduction in 
22 ; | manufacturing costs of high-speed circuits enabled DEC, among 

j others, "to enter the computer market with low-priced computers 
ZS : j of high performance". (PX 4829, p. 31.) 

24. 1 



-720- 



I 

2 

3 
4 

5 

6.1 

7 

3 

9- 
10 
11 
12 l 
13 
14 
15 
15 ; 

17| 

i 

is I 

19 

20 



b * PDP 6. The PDP 6 was first delivered by DEC in 1964. 
(Hindle, Tr. 7321.)*" DEC'S Annual Report for 1964 described the 
PDF 6 as "an expandable system which can start as a very basic 
configuration and grow through the addition of processor, memory 
and input-output options into a major computation facility equivalent 
to the largest commercial systems currently offered". (DX 13845, 
p. 6.) At the time, the PDP 6 was the largest of DEC'S computer 
systems, ranging in price from $350,000 to $750,000. DEC described 
it as "equivalent to the very large computers used by scientific 
laboratories". The PDP 6 was used in "large data processing assign- 
ments", including Brookhaven National Laboratory, the Rutgers 
University Physics Department, the Universities of Aachen and Bonn, 
Germany, the University of Western Australia, Lawrence Radiation 
Laboratory, United Aircraft Corporation, Applied Logic Corporation, 
Yale, MIT's Laboratory for Nuclear Science, the University of 
Rochester, Stanford University and the University of California at 
Berkeley. (DX 13845 , p. 3; DX 13846 , p. 8; DX 13847, p. 7.) 

Also, the PDP 6 was "designed for time-shared use" and 
'(DEC bid a multiprocessor version of it to MIT's Project MAC, one of 
the earliest and most important experiments in the use of time 



ij sharing. DEC was "in among the finishers" for this award, who 
| included: CDC, bidding a 6600; IBM, bidding a 360/50; GE, the 



22 
24; 
25 i 



* The PDP 6 was withdrawn in 196 7, when it was succeeded by the 
PDP 10 which "incorporates all the features of the earlier machine". 
(Hindle, Tr. 7321; PX 5026 (DX 13848, p. 3).) 



-721- 



Li 
i 

z\ 

l! 

t 

i 

! 

i 

7i 

8 | 

i 
Si 

-I 

u| 

i 

12 j 
13 
14 
15 
16 
17 

13 ; 

19 I 

20 ! 



winner of the contract with a modified 635; and RCA, with a 3301. 

A $1 million PDP 6 was chosen by Project MAC as a peripheral processor 

for the time-sharing system. (PX 2961, pp. 1, 3; DX 13845, p. 3.) 

An IBM Win and Loss Report for August 1964 reported additional 

« 

competition between the PDF 6 and the IBM 360/50. (PX 3630, p.. 6.)* 

At the Lawrence Berkeley Laboratory of the Atomic Energy 
Commission, a PDP 6 was bid against an IBM 360/70 and a CDC 6600, 
among others. A CDC 6600 was installed in January 1966. (DX 2992, 

p.- ID 

At Lawrence Livermore Laboratory , two PDP 6 ' s were acquired 
in 1965-*66 and used as control computers in the OCTOPUS network. 
Because Lawrence Livermore required a faster memory than DEC could 
supply, it solicited bids for add-on memory, receiving bids from 
five companies in addition to DEC, and awarding contracts to Lockheed 
and Ampex. (DX 4572.) That acquisition was an indication of 
things to come: there was later, primarily in the 1970s, substantial 
development of DEC plug-compatible equipment, in part reflecting 
the great popularity of DEC computer systems. (See, e.g., Hindle, 
Tr. 7422-23, 7444-45.) 

c. PDP 8. The PDP 6 procurements were prestigious, but 



•I 
^ :j DEC's financial growth was more affected by the PDP 8 series of 

21 ;| • 

li 

22 :; 

| * The United Aircraft Research Laboratories offered computation 

22 i services with equipment consisting of an IBM 360/50 system, a 

;! Univac 1108 system and a "DEC PDP -6 Time-sharing system" including 

24 it "a paper tape reader, and both drum and magnetic tape auxiliary 
;■ storage" and accessible "through most standard terminal units". 

25 ; (DX 7506, p. 44.) 

i 

! -722- 



I 

2 
Z 

4- 
5 

a 

7 
3 ! 

»i 
u | 

12 ( 

i 

u ! 

I 

i 
IS j 

17 j 

13 ; 

is- ;i 



computer systems — one of DEC ' s most successful. The PDP 8 was 
first introduced in 1964 as a replacement for the PDP 5. Due to 
the use of integrated circuits , the PDP 8 was four times faster 
than the PDP 5 at two-thirds the price. (Handle, Tr. 7321; DX 
13845/ p. 3; DX 13846/ p* 3.) Using the price/performance improve- 
ments made possible by the "FLIP-CHIP circuit modules and automated 
l| production techniques", the PDP 8 opened new market opportunities 
for DEC (such as typesetting) and expanded its base of scientific- 
oriented users. (DX 13846 , p. 3; DX 13847 , p. 8.) The PDP 8 was 
offered with "disk storage units, terminals, tape units, line 
printers, cathode-ray tube display units". (Hindle, Tr. 7334.) 

Perlis testified that the PDP 8 "generalized very nicely 
to other machines and it itself gave birth to a whole line of 
offspring" (Perlis, Tr. 1877); the "parent" itself "received a 
remarkably immediate acceptance". (DX 13847, p. 8.) As a result 
of the greater than expected demand, DEC expanded its manufacturing 
facilities in 1965. By mid-1966, over 400 PDP 8's had been installed* 
(DX 13846 , p. 3; DX 13847 , p. 8.) 

The various other members of the PDP 8 family in existence 
at the time of Hindle 's testimony in November 1975 were introduced 



20 : 

j] from 1966 (the PDP 8S) through 1974 or 1975 (the PDP 8A) with the 



21 '« 



j; later members of the family still in delivery. (Hindle, Tr. 7322.) 



22 
24 

25 



After the original introduction of the PDP 8, each new member of 
j the PDP 8 family was introduced at a lower price because of "changes 
in manufacturing technology, semiconductor prices, and peripheral 

-723- 



prices as purchased from our vendors". (Hindle, Tr. 7347-48.)* 
The systems of the PDP 8 family were software-compatible (Hindle, 
Tr. 7421) with DEC providing three "general purpose operating 
systems" for them. (Hindle { Tr, 7346-47.) It was "generally 
true"'/ although "not . . • . one hundred percent true", that the same 
peripherals could be used on all systems in the family. (Hindle, 
Tr. 7421-22.) 

First deliveries of the PDP 8 went to such organizations 
as Stanford Research Institute, Harvard Medical School, Massachusetts 
Institute of Technology and the University of Wisconsin. The PDP 8 
was also offered to the newspaper and book publishing fields for 
automatic typesetting (DX 13846, p. 6) and to laboratory, industrial 
and educational users. (Hindle, Tr. 7331.) 

In addition, approximately 30 to 50% of the PDP 8 family 
was sold OEM,** some as processors alone, some as processors plus 
memory, and some as total systems. (Hindle, Tr. 7330.) When sold 
OEM as a processor only, the purchaser would acquire "from another 
manufacturer the appropriate devices necessary to perform input or 
|! output functions", with either the customer or DEC providing "inter- 



im 
2 

3 

4j 

5 

a 

7\ 

3 

9 
10 
11 
12 
13 
14 
15 

17 I 
13 ! 

19 i 

h 

jj * Hindle testified that in pricing its products, DEC takes 
2 * ! i several factors into account: (i) the computer systems DEC believes 
;! will be competitive with the one being priced, (ii) conditions in 

22 ;j the segments of the market into which the product is expected to be 
i| sold, (iii) manufacturing and support costs, (iv) expected profit. 

23 : j (Tr. 7337-38.) 

24 ; ** OEMs are systems vendors or manufacturers which incorporated 

l| the PDP 8 processors into their systems or products. (Hindle, Tr. 
" ; | 7330-31.) DEC generally charged OEM buyers and end-user buyers the 
j same prices. (Hindle, Tr. 7 34 8.) 

" -724- 

.! 

■ i 

;i 



I 

2 

5 

el 

i 

7\ 

8 ! 
9-! 

t 

10 

I 

n ! 



facing services" for the input and output devices. (Hindle, Tr. 
7330-31.) Since 1964, these OEM purchasers have marketed PDP 8 
systems for business data processing. (Hindle, Tr. 7332.) 

Business data processing applications that are performed 
on PDP 8 computer systems include "invoicing, accounts payable, 
inventory control, order processing" and others. (Hindle, Tr. 
7389.) In addition, PDP 8' s have been used for "real time data 
collection from instruments", "to assist in the teaching process", 
in "industrial control applications", including "the automation of 
industrial process controls, the collection, analysis and reporting 
of quality control data, test data, material handling data", in 



J2, j "commercial typesetting applications" such as "copy editing, in 



hyphenization and justification" and "setting classified advertise- 
ments", in "data communications applications" such as "message 
switching, data multiplexing, data concentration" or "front end 
processing". Different users use the PDP 8 to perform different 
applications and in some cases the same user might use the same PDP 
8 computer system to do, for example, both business data processing 
and industrial control applications. (Hindle, Tr. 7389-91.) 

DEC itself, prior to 1972, had largely been "unsuccessful" 



23 
24 
25 



tions because it had not worked "on the packaging to make the 
product suitable and attractive to the business data processing 
customers". (Hindle, Tr. 7489-90.) Then, in 1972, DEC introduced 
the Datasystem 300, an adaptation of the PDP 8 specifically designed 



-725- 



L j for business data processing. The "primary difference" between the 

i 

Z-' { PDP 8 and the DEC Datasystem 300 was that the latter was packaged 

3 [ in a different type of console and used a business-oriented language 

4 1 called DIBOlr. The Datasystem 300 was offered with the same peripherals 

i • ' 

f j as the PDF 8.. (Hindle, Tr. 7333-34.) 

i 

I 

$ I During the period prior to 1975, the PDP 8 competed with 

i 

7 I IBM's 360 and 370 computer systems when used as part of larger DEC 

3_ ; computer systems like the PDP 10 and singly against the IBM System/3, 

g. | System/32 and 1130. (Hindle, Tr. 7341, 7442; PX 377-A.)* IBM r s 

John Akers recalled losing to a PDP in 1966, when he bid a 360/20 

for a typesetting application at Worcester Telegram. (Akers, Tr. 

j^ | 96713.) The PDP 8 also had the ability to perform terminal or 

12 i input/output applications as part of a computer system with an IBM 

! 

^j GPU or some other manufacturer's CPU. (Hindle, Tr. 7394.) DEC 

i 

je ! considered the products of IBM when setting the price for the 

i 

.- <\ Datasystem 300. (Hindle, Tr. 7338-39, see Tr. 7341.) 

1T ; By any standards, the PDP 8 was a successful and significant 

product. At the time of Hindle 's testimony in late 1975, between 



10 

11 



1S „ 
20 ; 



30,000 and 40,000 PDP 8's had been sold (Hindle, Tr. 7329), and the 



* Hindle testified that in drawing up PX 377-A (which was offered 
in evidence "for an illustrative purpose" and "to assist Mr. Hindle 
(j in testifying about each of the products listed thereon" (Tr. 7320- 
2- ij 21)) and in testifying to competition generally, he meant "one for 

j| one competition", i.e. , a situation in which both DEC and DEC's 
^ : competitor will bid for a product with the same price and the same 

|j performance to do the same job or where DEC "would bid two or three 
2-" ;j of our products to compete with one of the products from the other 
company". (Tr. 7414-17.) 



21 



25 



-726- 



2 

Z j 

a 

7 

3 

9 
10 
U 
12 
"13 



PDP 8 was being marketed "to a wider variety of customers", including 
"communications customers" and "business data processing customers". 
(Kindle, Tr. 7331-32.) 

<*• B®P 10'- In 1967, DEC introduced the PDP 10.* The 
PDP 10 was the first of a family of computer systems later marketed 
as the "DEC System 10". (Kindle, Tr. 7324.) Kindle testified that 
the PDP 10 could do "all of the different kinds of applications 
that, are performed by the PDP 8" as well as additional applications 
characterized by DEC as "non-business computation applications", 
such as the manipulation and analysis of "scientific, engineering, 
or numerical data".. (Tr» 7391-92.) The PDP 10 was "introduced to 
serve . . . laboratory users, industrial users, education users, a 
class of users we call the data service industry. . . . This is a 
• . f class of computer users who purchase— or lease or rent — computer 
.- jl equipment and then offer various kinds of services to clients." 

^m* L 

i (Kindle, Tr. 7359.) When introduced, it could have been purchased 
"for prices in the vicinity of $450,000, all the way up to configura- 
tions which would be a million and a half dollars". (Hindle, Tr. 
7359.) It was not actively marketed on an OEM basis, but generally 

i 

was sold directly to end users. (Hindle, Tr. 7358.) ! 

: ! 

i 

21 \ ~ | 

j * Using his definition of "minicomputer" (a computer system in j 

22 : | which "the smallest available configuration could be configured for j 
; j less than $50,000" (Tr. 7453)), Hindle classified all DEC computer j 

23 j systems introduced during the 1960s as "minicomputers" except for ; 
j the PDP 10 , which was too large and too expensive to meet that 

24. '"} definition. (Tr. 7325-27, 7358-59.) ! 

i i 

25 ; ■ ■ j 

! -727- 



17 j 
IS ; 

19 ; 

2Q : 



2 I 

i 
I 
I 

*l 

i 

5 I 
» 

i 

6 j 

7 | 

a! 

* I 

10-1 

11 

12 

13 

L4 

15 

16 

17 

IS 



In 1969, DEC described the PDP 10 as follows: 

"serving business, industry, and science in a multitude of 
installations throughout the world- They keep track of bubble 
chamber events in physics laboratories, analyze blood chromosomes, 
work in banks, teach in high schools and universities, and per- 
form a myriad of other, tasks. New applications are constantly 
appearing and current applications steadily grow. Customers find 
new approaches, add new equipment, develop more software. 
Systems designed solely for real-time tasks often expand to 
include program development or business data processing. The 
applications described here demonstrate the PDP-10's inherent 
flexibility." (DX 519-B, p. 7.) 

The PDP 10 competed on a "one for one basis" with the IBM 

360 r 370 and System/3 computer systems. (Hindle, Tr. 7442; PX 377- 

A.)* In establishing the price of the PDP 10, DEC looked at the IBM 

360 series, specifically the Models 30, 40 and 50, as well as systems 

offered by Honeywell, General Electric and Scientific Data Systems. 

(Hindle, Tr. 7361-62.) 

The DECsystem 10 was announced in 1971 as a family of 

systems "spanning virtually the entire large-scale range" and was 

based upon the PDP 10 processors.** (DX 512, p. 1; DX 522, p. 3.) 

(The DECsystem 10 is discussed in the section of this narrative deal- ■ 



.- :i * This competition is demonstrated by the procurements of govern- 
i| ment agencies. For example, at the Atomic Energy Commission, a PDP 10 

2 i| was proposed in competition with an IBM 360/44 and a SEL 810A. The | 

;■! SEL 810A was selected and installed in December 1968. Earlier that | 

-^ ij year, the Commission selected a Sigma 7 bid against an IBM 360/50 and j 

: a PDP 10. On another occasion, a PDP 10 was successfully bid against j 

| a 360/50 at the Commission and installed in late 1969. The Depart- j 

^ jj ment of Health, Education and Welfare chose a PDP 10 over a proposed j 

360/40 in 1969. (DX 2992, pp. 73, 86, 118, 858.) I 



23 • 

24 3 
25 



** The DECsystem 10 peripherals embodied some improvements compared 
to the peripherals offered with the PDP 10. In particular, there 
!| was a better quality printer and a disk drive with a removable disk 
'i pack. (Hindle, Tr. 7362-63.) 

;| -728- 

u 

l\ 

i 
it 



L 
Z 

z 

4- 
5 
a 



ing with the 1970s.) Both the PDP 10 and the DECsystem 10 were 
highly successful. By 1974, over 400 such systems had been installed. 
(Kindle, Tr* 7380; DX 525, p.. 10.) 

e * ?PP 15 » The PDP 15, introduced in 1969 (Hindle, Tr. 
7323) , was originally "marketed to laboratory users , industrial 
users, education users", but with the addition of another programming 



7; language ("MUMPS")— a "business-oriented language" — it was also 



3 

10 
11 
12 

13 



marketed for business data processing applications starting in the 
early seventies. (Hindle, Tr. 7377.) "Representative applications" 
performed by the PDP 15 were "nonbusiness computation applications, 
business data processing applications, real time data collection and 
instructional computing applications as well as industrial control 
applications". (Hindle, Tr. 7441; PX 377-A.) Hindle testified that 
the PDP 15 competed "on a one for one basis" with the IBM 1130, 1800, 
£5 j System/7, 360 and 370 computer systems. (Tr. 7441; PX 377-A.) By 
♦ fi ' late 1975, between 800 and 1,200 PDP 15s had been installed. (Hindle, 
y- j Tr. 7380.) 



e ., f. PDP 11. As with the PDP 8, the PDP 11, introduced in 

a j! 1970, was the designation for a family of computer systems. (Hindle, 
| Tr. 7323.) At the time of introduction DEC expected to market it "to 
j the entire group of users . . . described for the PDP 8, which would 
\ include laboratory users , education users , industrial users , engineer- 
„_ J ing users, [and] communications users". As with the PDP 8, between 30 

« i 

and 50 percent of the PDP lis were sold to OEM purchasers who wrote 



24 ) 
25 



applications programs and offered the PDP 11 for business data process- 

-729- 



ing applications.* . (Hindle, Tr. 7349-51.) But, DEC preferred to 
market the PDP 11 as. a system rather than just the processor. (Hindle, 
Tr. 7349.) In 1972:, the DEC Datasystem 500 series was introduced for 
marketing to business data processing customers. DEC took the PDP 11, 
added software capabilities, including BASIC, and put it in a "sepa- 
rate type of package" that looked different to the user. "But other 
than that, there were no significant differences ►" (Hindle, Tr. 7351- 
52, 7355-57.) 

Hindle testified that the PDP 11/15 and 11/20, smaller 
members of the PDP 11 family r competed on a one-for-one basis with the 
IBM 1130, System 3 and System 7 computers, and also with the 360 and 
370 in configurations which included the PDP 10. (Tr. 7441 t PX 377- 
A. ) Larger members of the PDP 11 family, like the 11/45 and 11/70, 
competed on a one-for-one basis with IBM 360 and 370, as well as 
System 3, System 7 and 1130.** (Tr. 7414-15; PX 377-A.) Like the PDP 8, 
the smaller PDP 11 computer systems might "be used to perform terminal 
or input/output applications" as part of computer systems whose main 



I 

V 

7 

5 

7 \ 
3 
3 

a 

1! 

2 

3 
.4 

J 1 

.3] 

;| "An OEM customer of ours . . . would then hire and train 

^ ,j and use programmers, who would write applications programs for 
;i the PDP 11 system for that particular problem application that 
;j he had identified as a marketing segment, and then would 
jj proceed to sell the combined PDP 11 system with the application 
programs that he had designed to the end user." (Tr. 7351.) 



* Hindle described the OEM marketing as follows : 



?*r 



' » 



23 H ** The PDP 11/45 and PDP 11/70 were announced in 1971 and 1975, 



respectively. DEC does not consider the PDP 11/70 to be a "mini- 
2A 1 computer". (Hindle, Tr. 7323, 7325-27.) Even some PDP ll/45s have 

i been configured into systems with prices as high as $250,000. 
25 j (Hindle, Tr. 7456.) 



-730- 



L 
Z 

3 

4 

5 
7 
3 

10 

II 
IZ 
13 



7 ^ p 1970 DEC had "introduced many new peripherals including those of our 



CPU was manufactured by IBM or by manufacturers other than IBM or 
DEC. (Hindle, Tr. 7394.) 

In. pricing the products in the DEC DataSystem 500 series, 
DEC looked at the prices of the IBM System 3, System 7 , System 32, 360 
and 370 systems.. (Hindle, Tr. 7354.) 

The PDP 11 family has been a highly successful and signifi- 
cant product line, but its story, like that of the DECsystem 10, 
unfolds in the 1970s. 

g. Peripherals and Software. The rapid proliferation of 
DEC'S product line during the 1960s extended to peripheral and soft- 
ware ' of fe rings as well. Although as late as 1969 Memorex was market- 
ing disk file products to DEC on an OEM basis (Spitters, Tr. 42067-68) 
with 1969 sales of approximately $5,000,000 (Spitters, Tr. 42072), by 



15 

15 _ 
17 | 

is ; 

19 

20 ;| 

21 | 
22 



23 ! 



24 
25 



own internal design and manufacture, such as disks, paper tape, 
| t DECtape, display systems, and real time interface equipment". (DX 
517, p. 2.) DEC's 1970 Annual Report proclaimed that "[i]n order to • 
expand the capabilities of its computers, DEC provides a wide range of 
I peripheral equipment", including large magnetic tape systems, storage 
drums, teletypes, high speed paper tape readers, card readers and 
punches, line printers, incremental plotters, digital-to-analog 
converters and various controllers. (DX 511, p. 10.) 

DEC had also worked on software, introducing new software 
features " [w] ith each mainframe that is a new version of a previous 
machine". (DX 517, p. 2.) It had provided "DIBOL" , a "business 



-731- 



1 

2 
3 
4* 
5 

6 
7 
3 

10 

11 



oriented language " for the PDP 8, and added COBOL to the PDP 10. (DX 
517, pp. 3, 5..) 

in 1967 it developed application packages called " Computer - 
packs" which from?: 1967 to 1$69 were marketed together with the DEC 
hardware at no separate charge. (Hindle, Tr. 7426.) The "Computer- 
pack" was actually a complete turnkey system for users desiring 
systems that required a minimum of programming and computer experience 
DEC merely added an application package to a PDP 8 and marketed 
the result as a "Computerpack" . For example, the "Quiokpoint-8" was 
offered for numerical control tape preparation; the "Communic-8" was 
offered for data communication applications; the "Time-Shared- 8" was 



t5 j offered for general purpose time-sharing applications; and the "LAB— 8" 



13 



15 



was offered for nuclear magnetic resonance spectroscopy applications. 



l±\ (DX 6868, pp. 7-10; DX 10776, pp. 8-13.) 



h. Competition. DEC ' s approach to the market was different 



^g j| than that of IBM. DEC in the 1960s offered fast, inexpensive hardware 



17 
13 



with less versatile and generalized software and service than that 
offered by IBM. As shown in PX 3 77 -A, DEC marketed most of its 

• a :! machines announced in the 1960s to "experienced" and "moderately 

'I 
2q ! j experienced" users . Perlis described the PDP systems as they were 

2«r ij perceived in the university environment: "[I]t was generally felt 

-«. i! that . . . PDP systems . . . for delivering the same amount of work, 

i 

-- |j were cheaper than the I3M systems." He estimated that the PDP 10 was 

_ A I about 20% cheaper than a 360/50 because of the "attendant staff of 

« ' operators, people to handle the variety of software that is used . . . 

-732- 

i 
■I 



and so forth" associated with the IBM 360 Model 50 while PDP 10' s were 
operated "without any staff whatsoever in attendance on the machine 
during its period of operation, which runs 24 hours a day, seven days 
a week".* (Tr„ 1976-77.) In this respect, the bundle of services 
associated with the IBM 360 line provided an opportunity for DEC to 
obtain a price advantage with users who did not want or need those 
services. Thus, as is illustrated by the discussion below, DEC 
offered hardware and software more tailored than the generalized 
System/360 to enable users to perform one or a few applications in a 
decentralized way rather than on a central IBM computer. 

Competition from DEC was felt within IBM in the 1960s. 
Wright, who was a Director of Marketing in IBM's Data Processing 
Division in the 1960s, included DEC on his list of "principal competi- 
tors" during the 1964 to 1969 time period. (Wright, Tr. 12993.) 
Similarly, when Rooney was employed by IBM in the mid-1960s as a 
Branch Manager in New York, DEC was competing in the "marketplace" for 
the "manufacture and marketing of systems for commercial or scientific 
usage". (Rooney, Tr. 11733.) Akers recalled meeting DEC in three 
different situations in which he was personally involved in the 1960s 



Tr. 96713-14.) He studied DEC both as a salesman in Vermont and a 



22 
23 
24 
25 



* Of course, with IBM "the user receives an enormous amount of 
service, an enormous amount of software, very good maintenance and for 
many users that is well worth the 20 per cent difference". (Perlis, 
Tr. 1978.) 



-733- 



1 

2 
Z 

5 

S 

7 

3 

9 

10 

11 
12 
13 
14 
15 
IS 
17 
IS 



marketing manager in Boston (Tr. 96679) and, when in the New York 
Media branch office, found that IBM had "a good deal of competition 
from the Digital Equipment Corporation". (Tr. 96680.) 

The competition between DEC and IBM was not only on a "one~- 
for-one" basis. As Hindle explained, " [i]t would be possible that in 
a given computer application a customer could choose one powerful 
machine to do the job or could choose several less powerful machines 
and decentralize the job. In that type of situation we would have one 
machine competing with several machines from a different manufacturer." 
Such competition would arise, for example, because " [considering the 
total system cost of both software and hardware, a distributed network 
of smaller computers can often be a cost-effective alternative to the 
single, centralized computer". Alternatively, as Hindle said, ,r [i]t 
is possible to have several smaller computer systems which are not 
interconnected electronically" competing with a single larger computer 
system. (Tr. 7415-17.) 

Similarly, the IBM Commercial Analysis Department described 
this competition in the Quarterly Product Line Assessment for the 
1$ ;| first quarter of 1970: "Mini-computers affect IBM's business poten- 
20 : j tial by implementing one application out of several possible applica- 
2x I! tions in a prospect's business." And, according to the report, the 

22 *•; application selected for the minicomputer was frequently the applica- 

23 ! tion having the greatest economic justification. The off-loading of 

24 ; ! that application could eliminate the opportunity for IBM to supply the 
-25 customer with a "larger and more comprehensive computer installation". 

-734- 



L 
Z 

z 

4- 

5 

**■ 
s > 

7J 

i 

al 

9- 
10 

II 
EL 
13 

14- 
15 
IS 
17 
IS 

»i 

20 :! 

* 'I 

\ 

22) 
1 

23 J 



4^ | 



(PX 2567, p. 186.) The report also commented on the success of 

minicomputers : 

"Hani-computers have established a substantial base and 
continue to widen their base each year. Digital Equipment 
Corporation is now ranked #3 in total CPU's representing 
about 5,600. units (8%)- out of the 70,000 total domestic 
CPU's reported for year end 1969- 'by Diebold." ( Id. ) 

Other companies also met DEC as competition in the 
1960s. Palevsky testified that DEC competed with SI>S ' s Sigma 
Series. (Tr. 322S-29. ) Honeywell management believed that Honeywell 
systems competed with systems from DEC as well as those from General 
Electric and Hewlett-Packard.* (Binger, Tr. 4593-94.) 

DEC entered the seventies a large and profitable company 
with a successful and popular product line. Its 1970 fiscal year 
revenues were $135.4 million with income before taxes of $25.5 
million. (DX 511, p. 1.) It had some 500 computers installed in 
the Federal government, or almost 10% of the total number, making 
it the third ranked supplier in terms of numbers of computers 
(behind IBM and Univac) . (DX 924, p. 6.) But, despite the 
impressiveness of those indicators of success, they were but 
small fractions of the DEC that was to emerge in the next decade. 



* Spangle included DEC as well as Sperry Rand, NCR, Burroughs 



25 ] and CDC in a list of Honeywell's competitors. (Spangle, Tr. 
4933-34.) 

; i 

;! -735- 






3 



I * 

T 



49.. AT&T .. Despite the continuing restrictions of its 1956 
Consent Decree, AT&T expanded its offerings of computer-related 
products and services during the 1964-1969 period. Its U.S. EDP 
: revenues, a*s a result, grew from- $125.6 million in 1964 to $477.75 
fj| million by 1969- ' (DX 8224, p. 133.) 

As it did in the 1950s, AT&T competed in the computer 
industry in at least two ways during the 1960s. The first involved 
Western Electric ' s manufacture and marketing to the Bell System 
operating companies of stored program controlled electronic switching 
systems and automatic intercept systems. Because of the Bell System's 
enormous size and the fact that the Bell operating companies are free 
to and do in fact buy EDP products and services from non-Bell 
affiliated companies, this is a very important source of business to 
computer vendors who vie with Western Electric for the business of 
the Bell System. (See DX 5945, Dunnaville, pp. 6-8 and discussion 
below.) During the 1960s equipment developed and manufactured, by 
AT&T competed for- the business of the Bell System with the equipment 
of other EDP vendors, including IBM. (Id.)* 



*3 



17 

19 ii 



! The second form of competition is AT&am